Merge branch 'dev' into 230626-optimize-output-frame-resize

.devcontainer/post_create.sh

@@ -14,6 +14,11 @@ curl -L https://api.github.com/meta | jq -r '.ssh_keys | .[]' | \
 sudo mkdir -p /media/frigate
 sudo chown -R "$(id -u):$(id -g)" /media/frigate
 
+# When started as a service, LIBAVFORMAT_VERSION_MAJOR is defined in the
+# s6 service file. For dev, where frigate is started from an interactive
+# shell, we define it in .bashrc instead.
+echo 'export LIBAVFORMAT_VERSION_MAJOR=$(ffmpeg -version | grep -Po "libavformat\W+\K\d+")' >> $HOME/.bashrc
+
 make version
 
 cd web

Dockerfile

@@ -18,10 +18,13 @@ WORKDIR /rootfs
 
 FROM base AS nginx
 ARG DEBIAN_FRONTEND
+ENV CCACHE_DIR /root/.ccache
+ENV CCACHE_MAXSIZE 2G
 
 # bind /var/cache/apt to tmpfs to speed up nginx build
 RUN --mount=type=tmpfs,target=/tmp --mount=type=tmpfs,target=/var/cache/apt \
     --mount=type=bind,source=docker/build_nginx.sh,target=/deps/build_nginx.sh \
+    --mount=type=cache,target=/root/.ccache \
     /deps/build_nginx.sh
 
 FROM wget AS go2rtc
@@ -61,14 +64,16 @@ RUN mkdir /models \
 FROM wget as libusb-build
 ARG TARGETARCH
 ARG DEBIAN_FRONTEND
+ENV CCACHE_DIR /root/.ccache
+ENV CCACHE_MAXSIZE 2G
 
 # Build libUSB without udev.  Needed for Openvino NCS2 support
 WORKDIR /opt
-RUN apt-get update && apt-get install -y unzip build-essential automake libtool
+RUN apt-get update && apt-get install -y unzip build-essential automake libtool ccache
-RUN wget -q https://github.com/libusb/libusb/archive/v1.0.25.zip -O v1.0.25.zip && \
+RUN --mount=type=cache,target=/root/.ccache wget -q https://github.com/libusb/libusb/archive/v1.0.25.zip -O v1.0.25.zip && \
     unzip v1.0.25.zip && cd libusb-1.0.25 && \
     ./bootstrap.sh && \
-    ./configure --disable-udev --enable-shared && \
+    ./configure CC='ccache gcc' CCX='ccache g++' --disable-udev --enable-shared && \
     make -j $(nproc --all)
 RUN apt-get update && \
     apt-get install -y --no-install-recommends libusb-1.0-0-dev && \
@@ -93,7 +98,9 @@ COPY labelmap.txt .
 COPY --from=ov-converter /models/public/ssdlite_mobilenet_v2/FP16 openvino-model
 RUN wget -q https://github.com/openvinotoolkit/open_model_zoo/raw/master/data/dataset_classes/coco_91cl_bkgr.txt -O openvino-model/coco_91cl_bkgr.txt && \
     sed -i 's/truck/car/g' openvino-model/coco_91cl_bkgr.txt
-
+# Get Audio Model and labels
+RUN wget -qO cpu_audio_model.tflite https://tfhub.dev/google/lite-model/yamnet/classification/tflite/1?lite-format=tflite
+COPY audio-labelmap.txt .
 
 
 FROM wget AS s6-overlay
@@ -127,7 +134,9 @@ RUN apt-get -qq update \
     libtbb2 libtbb-dev libdc1394-22-dev libopenexr-dev \
     libgstreamer-plugins-base1.0-dev libgstreamer1.0-dev \
     # scipy dependencies
-    gcc gfortran libopenblas-dev liblapack-dev && \
+    gcc gfortran libopenblas-dev liblapack-dev \
+    # faster-fifo dependencies
+    g++ cython3 && \
     rm -rf /var/lib/apt/lists/*
 
 RUN wget -q https://bootstrap.pypa.io/get-pip.py -O get-pip.py \

audio-labelmap.txt

@@ -0,0 +1,521 @@
+speech
+speech
+speech
+speech
+babbling
+speech
+yell
+bellow
+whoop
+yell
+yell
+yell
+whispering
+laughter
+laughter
+laughter
+snicker
+laughter
+laughter
+crying
+crying
+crying
+yell
+sigh
+singing
+choir
+sodeling
+chant
+mantra
+child_singing
+synthetic_singing
+rapping
+humming
+groan
+grunt
+whistling
+breathing
+wheeze
+snoring
+gasp
+pant
+snort
+cough
+throat_clearing
+sneeze
+sniff
+run
+shuffle
+footsteps
+chewing
+biting
+gargling
+stomach_rumble
+burping
+hiccup
+fart
+hands
+finger_snapping
+clapping
+heartbeat
+heart_murmur
+cheering
+applause
+chatter
+crowd
+speech
+children_playing
+animal
+pets
+dog
+bark
+yip
+howl
+bow-wow
+growling
+whimper_dog
+cat
+purr
+meow
+hiss
+caterwaul
+livestock
+horse
+clip-clop
+neigh
+cattle
+moo
+cowbell
+pig
+oink
+goat
+bleat
+sheep
+fowl
+chicken
+cluck
+cock-a-doodle-doo
+turkey
+gobble
+duck
+quack
+goose
+honk
+wild_animals
+roaring_cats
+roar
+bird
+chird
+chirp
+squawk
+pigeon
+coo
+crow
+caw
+owl
+hoot
+flapping_wings
+dogs
+rats
+mouse
+patter
+insect
+cricket
+mosquito
+fly
+buzz
+buzz
+frog
+croak
+snake
+rattle
+whale_vocalization
+music
+musical_instrument
+plucked_string_instrument
+guitar
+electric_guitar
+bass_guitar
+acoustic_guitar
+steel_guitar
+tapping
+strum
+banjo
+sitar
+mandolin
+zither
+ukulele
+keyboard
+piano
+electric_piano
+organ
+electronic_organ
+hammond_organ
+synthesizer
+sampler
+harpsichord
+percussion
+drum_kit
+drum_machine
+drum
+snare_drum
+rimshot
+drum_roll
+bass_drum
+timpani
+tabla
+cymbal
+hi-hat
+wood_block
+tambourine
+rattle
+maraca
+gong
+tubular_bells
+mallet_percussion
+marimba
+glockenspiel
+vibraphone
+steelpan
+orchestra
+brass_instrument
+french_horn
+trumpet
+trombone
+bowed_string_instrument
+string_section
+violin
+pizzicato
+cello
+double_bass
+wind_instrument
+flute
+saxophone
+clarinet
+harp
+bell
+church_bell
+jingle_bell
+bicycle_bell
+tuning_fork
+chime
+wind_chime
+change_ringing
+harmonica
+accordion
+bagpipes
+didgeridoo
+shofar
+theremin
+singing_bowl
+scratching
+pop_music
+hip_hop_music
+beatboxing
+rock_music
+heavy_metal
+punk_rock
+grunge
+progressive_rock
+rock_and_roll
+psychedelic_rock
+rhythm_and_blues
+soul_music
+reggae
+country
+swing_music
+bluegrass
+funk
+folk_music
+middle_eastern_music
+jazz
+disco
+classical_music
+opera
+electronic_music
+house_music
+techno
+dubstep
+drum_and_bass
+electronica
+electronic_dance_music
+ambient_music
+trance_music
+music_of_latin_america
+salsa_music
+flamenco
+blues
+music_for_children
+new-age_music
+vocal_music
+a_capella
+music_of_africa
+afrobeat
+christian_music
+gospel_music
+music_of_asia
+carnatic_music
+music_of_bollywood
+ska
+traditional_music
+independent_music
+song
+background_music
+theme_music
+jingle
+soundtrack_music
+lullaby
+video_game_music
+christmas_music
+dance_music
+wedding_music
+happy_music
+sad_music
+tender_music
+exciting_music
+angry_music
+scary_music
+wind
+rustling_leaves
+wind_noise
+thunderstorm
+thunder
+water
+rain
+raindrop
+rain_on_surface
+stream
+waterfall
+ocean
+waves
+steam
+gurgling
+fire
+crackle
+vehicle
+boat
+sailboat
+rowboat
+motorboat
+ship
+motor_vehicle
+car
+honk
+toot
+car_alarm
+power_windows
+skidding
+tire_squeal
+car_passing_by
+race_car
+truck
+air_brake
+air_horn
+reversing_beeps
+ice_cream_truck
+bus
+emergency_vehicle
+police_car
+ambulance
+fire_engine
+motorcycle
+traffic_noise
+rail_transport
+train
+train_whistle
+train_horn
+railroad_car
+train_wheels_squealing
+subway
+aircraft
+aircraft_engine
+jet_engine
+propeller
+helicopter
+fixed-wing_aircraft
+bicycle
+skateboard
+engine
+light_engine
+dental_drill's_drill
+lawn_mower
+chainsaw
+medium_engine
+heavy_engine
+engine_knocking
+engine_starting
+idling
+accelerating
+door
+doorbell
+ding-dong
+sliding_door
+slam
+knock
+tap
+squeak
+cupboard_open_or_close
+drawer_open_or_close
+dishes
+cutlery
+chopping
+frying
+microwave_oven
+blender
+water_tap
+sink
+bathtub
+hair_dryer
+toilet_flush
+toothbrush
+electric_toothbrush
+vacuum_cleaner
+zipper
+keys_jangling
+coin
+scissors
+electric_shaver
+shuffling_cards
+typing
+typewriter
+computer_keyboard
+writing
+alarm
+telephone
+telephone_bell_ringing
+ringtone
+telephone_dialing
+dial_tone
+busy_signal
+alarm_clock
+siren
+civil_defense_siren
+buzzer
+smoke_detector
+fire_alarm
+foghorn
+whistle
+steam_whistle
+mechanisms
+ratchet
+clock
+tick
+tick-tock
+gears
+pulleys
+sewing_machine
+mechanical_fan
+air_conditioning
+cash_register
+printer
+camera
+single-lens_reflex_camera
+tools
+hammer
+jackhammer
+sawing
+filing
+sanding
+power_tool
+drill
+explosion
+gunshot
+machine_gun
+fusillade
+artillery_fire
+cap_gun
+fireworks
+firecracker
+burst
+eruption
+boom
+wood
+chop
+splinter
+crack
+glass
+chink
+shatter
+liquid
+splash
+slosh
+squish
+drip
+pour
+trickle
+gush
+fill
+spray
+pump
+stir
+boiling
+sonar
+arrow
+whoosh
+thump
+thunk
+electronic_tuner
+effects_unit
+chorus_effect
+basketball_bounce
+bang
+slap
+whack
+smash
+breaking
+bouncing
+whip
+flap
+scratch
+scrape
+rub
+roll
+crushing
+crumpling
+tearing
+beep
+ping
+ding
+clang
+squeal
+creak
+rustle
+whir
+clatter
+sizzle
+clicking
+clickety-clack
+rumble
+plop
+jingle
+hum
+zing
+boing
+crunch
+silence
+sine_wave
+harmonic
+chirp_tone
+sound_effect
+pulse
+inside
+inside
+inside
+outside
+outside
+reverberation
+echo
+noise
+environmental_noise
+static
+mains_hum
+distortion
+sidetone
+cacophony
+white_noise
+pink_noise
+throbbing
+vibration
+television
+radio
+field_recording

benchmark_motion.py

@@ -12,16 +12,32 @@ from frigate.util import create_mask
 # get info on the video
 # cap = cv2.VideoCapture("debug/front_cam_2023_05_23_08_41__2023_05_23_08_43.mp4")
 # cap = cv2.VideoCapture("debug/motion_test_clips/rain_1.mp4")
-cap = cv2.VideoCapture("debug/motion_test_clips/ir_off.mp4")
+cap = cv2.VideoCapture("debug/motion_test_clips/lawn_mower_night_1.mp4")
 # cap = cv2.VideoCapture("airport.mp4")
 width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
 height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
 fps = cap.get(cv2.CAP_PROP_FPS)
 frame_shape = (height, width, 3)
+# Nick back:
+# "1280,0,1280,316,1170,216,1146,126,1016,127,979,82,839,0",
+# "310,350,300,402,224,405,241,354",
+# "378,0,375,26,0,23,0,0",
+# Front door:
+# "1080,0,1080,339,1010,280,1020,169,777,163,452,170,318,299,191,365,186,417,139,470,108,516,40,530,0,514,0,0",
+# "336,833,438,1024,346,1093,103,1052,24,814",
+# Back
+# "1855,0,1851,100,1289,96,1105,161,1045,119,890,121,890,0",
+# "505,95,506,138,388,153,384,114",
+# "689,72,689,122,549,134,547,89",
+# "261,134,264,176,169,195,167,158",
+# "145,159,146,202,70,220,65,183",
 
 mask = create_mask(
     (height, width),
-    [],
+    [
+        "1080,0,1080,339,1010,280,1020,169,777,163,452,170,318,299,191,365,186,417,139,470,108,516,40,530,0,514,0,0",
+        "336,833,438,1024,346,1093,103,1052,24,814",
+    ],
 )
 
 # create the motion config
@@ -29,7 +45,7 @@ motion_config_1 = MotionConfig()
 motion_config_1.mask = np.zeros((height, width), np.uint8)
 motion_config_1.mask[:] = mask
 # motion_config_1.improve_contrast = 1
-# motion_config_1.frame_height = 150
+motion_config_1.frame_height = 150
 # motion_config_1.frame_alpha = 0.02
 # motion_config_1.threshold = 30
 # motion_config_1.contour_area = 10
@@ -38,10 +54,11 @@ motion_config_2 = MotionConfig()
 motion_config_2.mask = np.zeros((height, width), np.uint8)
 motion_config_2.mask[:] = mask
 # motion_config_2.improve_contrast = 1
-# motion_config_2.frame_height = 150
+motion_config_2.frame_height = 150
 # motion_config_2.frame_alpha = 0.01
-# motion_config_2.threshold = 20
+motion_config_2.threshold = 20
 # motion_config.contour_area = 10
+
 save_images = True
 
 improved_motion_detector_1 = ImprovedMotionDetector(
@@ -52,8 +69,6 @@ improved_motion_detector_1 = ImprovedMotionDetector(
     threshold=mp.Value("i", motion_config_1.threshold),
     contour_area=mp.Value("i", motion_config_1.contour_area),
     name="default",
-    clipLimit=2.0,
-    tileGridSize=(8, 8),
 )
 improved_motion_detector_1.save_images = save_images
 

docker/build_nginx.sh

@@ -15,6 +15,10 @@ apt-get -yqq build-dep nginx
 
 apt-get -yqq install --no-install-recommends ca-certificates wget
 update-ca-certificates -f
+apt install -y ccache
+
+export PATH="/usr/lib/ccache:$PATH"
+
 mkdir /tmp/nginx
 wget -nv https://nginx.org/download/nginx-${NGINX_VERSION}.tar.gz
 tar -zxf nginx-${NGINX_VERSION}.tar.gz -C /tmp/nginx --strip-components=1
@@ -62,5 +66,5 @@ cd /tmp/nginx
     --add-module=../nginx-rtmp-module \
     --with-cc-opt="-O3 -Wno-error=implicit-fallthrough"
 
-make -j$(nproc) && make install
+make CC="ccache gcc" -j$(nproc) && make install
 rm -rf /usr/local/nginx/html /usr/local/nginx/conf/*.default

docker/rootfs/etc/s6-overlay/s6-rc.d/frigate/run

@@ -44,6 +44,7 @@ function migrate_db_path() {
 
 echo "[INFO] Preparing Frigate..."
 migrate_db_path
+export LIBAVFORMAT_VERSION_MAJOR=$(ffmpeg -version | grep -Po 'libavformat\W+\K\d+')
 
 echo "[INFO] Starting Frigate..."
 

docker/rootfs/usr/local/go2rtc/create_config.py

@@ -7,7 +7,7 @@ import sys
 import yaml
 
 sys.path.insert(0, "/opt/frigate")
-from frigate.const import BIRDSEYE_PIPE, BTBN_PATH  # noqa: E402
+from frigate.const import BIRDSEYE_PIPE  # noqa: E402
 from frigate.ffmpeg_presets import (  # noqa: E402
     parse_preset_hardware_acceleration_encode,
 )
@@ -71,7 +71,7 @@ elif go2rtc_config["rtsp"].get("default_query") is None:
     go2rtc_config["rtsp"]["default_query"] = "mp4"
 
 # need to replace ffmpeg command when using ffmpeg4
-if not os.path.exists(BTBN_PATH):
+if int(os.environ["LIBAVFORMAT_VERSION_MAJOR"]) < 59:
     if go2rtc_config.get("ffmpeg") is None:
         go2rtc_config["ffmpeg"] = {
             "rtsp": "-fflags nobuffer -flags low_delay -stimeout 5000000 -user_agent go2rtc/ffmpeg -rtsp_transport tcp -i {input}"

docs/docs/configuration/audio_detectors.md

@@ -0,0 +1,63 @@
+---
+id: audio_detectors
+title: Audio Detectors
+---
+
+Frigate provides a builtin audio detector which runs on the CPU. Compared to object detection in images, audio detection is a relatively lightweight operation so the only option is to run the detection on a CPU.
+
+## Configuration
+
+Audio events work by detecting a type of audio and creating an event, the event will end once the type of audio has not been heard for the configured amount of time. Audio events save a snapshot at the beginning of the event as well as recordings throughout the event. The recordings are retained using the configured recording retention.
+
+### Enabling Audio Events
+
+Audio events can be enabled for all cameras or only for specific cameras.
+
+```yaml
+
+audio: # <- enable audio events for all camera
+  enabled: True
+
+cameras:
+  front_camera:
+    ffmpeg:
+    ...
+    audio:
+      enabled: True # <- enable audio events for the front_camera
+```
+
+If you are using multiple streams then you must set the `audio` role on the stream that is going to be used for audio detection, this can be any stream but the stream must have audio included.
+
+:::note
+
+The ffmpeg process for capturing audio will be a separate connection to the camera along with the other roles assigned to the camera, for this reason it is recommended that the go2rtc restream is used for this purpose. See [the restream docs](/configuration/restream.md) for more information.
+
+:::
+
+```yaml
+cameras:
+  front_camera:
+    ffmpeg:
+      inputs:
+        - path: rtsp://.../main_stream
+          roles:
+            - record
+        - path: rtsp://.../sub_stream # <- this stream must have audio enabled
+          roles:
+            - audio
+            - detect
+```
+
+### Configuring Audio Events
+
+The included audio model has over 500 different types of audio that can be detected, many of which are not practical. By default `bark`, `speech`, `yell`, and `scream` are enabled but these can be customized.
+
+```yaml
+audio:
+  enabled: True
+  listen:
+    - bark
+    - scream
+    - speech
+    - yell
+```

docs/docs/configuration/index.md

@@ -138,6 +138,20 @@ model:
   labelmap:
     2: vehicle
 
+# Optional: Audio Events Configuration
+# NOTE: Can be overridden at the camera level
+audio:
+  # Optional: Enable audio events (default: shown below)
+  enabled: False
+  # Optional: Configure the amount of seconds without detected audio to end the event (default: shown below)
+  max_not_heard: 30
+  # Optional: Types of audio to listen for (default: shown below)
+  listen:
+    - bark
+    - scream
+    - speech
+    - yell
+
 # Optional: logger verbosity settings
 logger:
   # Optional: Default log verbosity (default: shown below)
@@ -189,6 +203,11 @@ ffmpeg:
     record: preset-record-generic
     # Optional: output args for rtmp streams (default: shown below)
     rtmp: preset-rtmp-generic
+  # Optional: Time in seconds to wait before ffmpeg retries connecting to the camera. (default: shown below)
+  # If set too low, frigate will retry a connection to the camera's stream too frequently, using up the limited streams some cameras can allow at once
+  # If set too high, then if a ffmpeg crash or camera stream timeout occurs, you could potentially lose up to a maximum of retry_interval second(s) of footage
+  # NOTE: this can be a useful setting for Wireless / Battery cameras to reduce how much footage is potentially lost during a connection timeout.
+  retry_interval: 10
 
 # Optional: Detect configuration
 # NOTE: Can be overridden at the camera level
@@ -275,7 +294,7 @@ motion:
   # Optional: The threshold passed to cv2.threshold to determine if a pixel is different enough to be counted as motion. (default: shown below)
   # Increasing this value will make motion detection less sensitive and decreasing it will make motion detection more sensitive.
   # The value should be between 1 and 255.
-  threshold: 20
+  threshold: 30
   # Optional: The percentage of the image used to detect lightning or other substantial changes where motion detection
   #           needs to recalibrate. (default: shown below)
   # Increasing this value will make motion detection more likely to consider lightning or ir mode changes as valid motion.
@@ -448,10 +467,11 @@ cameras:
         # Required: the path to the stream
         # NOTE: path may include environment variables, which must begin with 'FRIGATE_' and be referenced in {}
         - path: rtsp://viewer:{FRIGATE_RTSP_PASSWORD}@10.0.10.10:554/cam/realmonitor?channel=1&subtype=2
-          # Required: list of roles for this stream. valid values are: detect,record,rtmp
+          # Required: list of roles for this stream. valid values are: audio,detect,record,rtmp
-          # NOTICE: In addition to assigning the record and rtmp roles,
+          # NOTICE: In addition to assigning the audio, record, and rtmp roles,
           # they must also be enabled in the camera config.
           roles:
+            - audio
             - detect
             - record
             - rtmp

docs/docs/configuration/object_detectors.md

@@ -1,6 +1,6 @@
 ---
-id: detectors
+id: object_detectors
-title: Detectors
+title: Object Detectors
 ---
 
 Frigate provides the following builtin detector types: `cpu`, `edgetpu`, `openvino`, and `tensorrt`. By default, Frigate will use a single CPU detector. Other detectors may require additional configuration as described below. When using multiple detectors they will run in dedicated processes, but pull from a common queue of detection requests from across all cameras.

docs/docs/configuration/restream.md

@@ -67,6 +67,7 @@ cameras:
           roles:
             - record
             - detect
+            - audio # <- only necessary if audio detection is enabled
   http_cam:
     ffmpeg:
       output_args:
@@ -77,6 +78,7 @@ cameras:
           roles:
             - record
             - detect
+            - audio # <- only necessary if audio detection is enabled
 ```
 
 ### With Sub Stream
@@ -112,6 +114,7 @@ cameras:
         - path: rtsp://127.0.0.1:8554/rtsp_cam_sub # <--- the name here must match the name of the camera_sub in restream
           input_args: preset-rtsp-restream
           roles:
+            - audio # <- only necessary if audio detection is enabled
             - detect
   http_cam:
     ffmpeg:
@@ -125,6 +128,7 @@ cameras:
         - path: rtsp://127.0.0.1:8554/http_cam_sub # <--- the name here must match the name of the camera_sub in restream
           input_args: preset-rtsp-restream
           roles:
+            - audio # <- only necessary if audio detection is enabled
             - detect
 ```
 

docs/docs/frigate/hardware.md

@@ -50,7 +50,7 @@ The OpenVINO detector type is able to run on:
 - 6th Gen Intel Platforms and newer that have an iGPU
 - x86 & Arm64 hosts with VPU Hardware (ex: Intel NCS2)
 
-More information is available [in the detector docs](/configuration/detectors#openvino-detector)
+More information is available [in the detector docs](/configuration/object_detectors#openvino-detector)
 
 Inference speeds vary greatly depending on the CPU, GPU, or VPU used, some known examples are below:
 
@@ -72,7 +72,7 @@ Inference speeds vary greatly depending on the CPU, GPU, or VPU used, some known
 
 ### TensorRT
 
-The TensortRT detector is able to run on x86 hosts that have an Nvidia GPU which supports the 11.x series of CUDA libraries. The minimum driver version on the host system must be `>=450.80.02`. Also the GPU must support a Compute Capability of `5.0` or greater. This generally correlates to a Maxwell-era GPU or newer, check the [TensorRT docs for more info](/configuration/detectors#nvidia-tensorrt-detector).
+The TensortRT detector is able to run on x86 hosts that have an Nvidia GPU which supports the 11.x series of CUDA libraries. The minimum driver version on the host system must be `>=450.80.02`. Also the GPU must support a Compute Capability of `5.0` or greater. This generally correlates to a Maxwell-era GPU or newer, check the [TensorRT docs for more info](/configuration/object_detectors#nvidia-tensorrt-detector).
 
 Inference speeds will vary greatly depending on the GPU and the model used.
 `tiny` variants are faster than the equivalent non-tiny model, some known examples are below:

docs/docs/guides/getting_started.md

@@ -71,7 +71,7 @@ cameras:
       ...
 ```
 
-More details on available detectors can be found [here](../configuration/detectors.md).
+More details on available detectors can be found [here](../configuration/object_detectors.md).
 
 Restart Frigate and you should start seeing detections for `person`. If you want to track other objects, they will need to be added according to the [configuration file reference](../configuration/index.md#full-configuration-reference).
 

docs/docs/integrations/mqtt.md

@@ -109,11 +109,19 @@ Same data available at `/api/stats` published at a configurable interval.
 
 ### `frigate/<camera_name>/detect/set`
 
-Topic to turn detection for a camera on and off. Expected values are `ON` and `OFF`.
+Topic to turn object detection for a camera on and off. Expected values are `ON` and `OFF`.
 
 ### `frigate/<camera_name>/detect/state`
 
-Topic with current state of detection for a camera. Published values are `ON` and `OFF`.
+Topic with current state of object detection for a camera. Published values are `ON` and `OFF`.
+
+### `frigate/<camera_name>/audio/set`
+
+Topic to turn audio detection for a camera on and off. Expected values are `ON` and `OFF`.
+
+### `frigate/<camera_name>/audio/state`
+
+Topic with current state of audio detection for a camera. Published values are `ON` and `OFF`.
 
 ### `frigate/<camera_name>/recordings/set`
 
@@ -176,7 +184,7 @@ Topic to send PTZ commands to camera.
 
 | Command                | Description                                                                               |
 | ---------------------- | ----------------------------------------------------------------------------------------- |
-| `preset-<preset_name>` | send command to move to preset with name `<preset_name>`                                  |
+| `preset_<preset_name>` | send command to move to preset with name `<preset_name>`                                  |
 | `MOVE_<dir>`           | send command to continuously move in `<dir>`, possible values are [UP, DOWN, LEFT, RIGHT] |
 | `ZOOM_<dir>`           | send command to continuously zoom `<dir>`, possible values are [IN, OUT]                  |
 | `STOP`                 | send command to stop moving                                                               |

docs/sidebars.js

@@ -16,7 +16,8 @@ module.exports = {
     ],
     Configuration: [
       "configuration/index",
-      "configuration/detectors",
+      "configuration/object_detectors",
+      "configuration/audio_detectors",
       "configuration/cameras",
       "configuration/masks",
       "configuration/record",

frigate/app.py

@@ -6,12 +6,13 @@ import shutil
 import signal
 import sys
 import traceback
-from multiprocessing.queues import Queue
 from multiprocessing.synchronize import Event as MpEvent
 from types import FrameType
 from typing import Optional
 
+import faster_fifo as ff
 import psutil
+from faster_fifo import Queue
 from peewee_migrate import Router
 from playhouse.sqlite_ext import SqliteExtDatabase
 from playhouse.sqliteq import SqliteQueueDatabase
@@ -29,6 +30,7 @@ from frigate.const import (
     MODEL_CACHE_DIR,
     RECORD_DIR,
 )
+from frigate.events.audio import listen_to_audio
 from frigate.events.cleanup import EventCleanup
 from frigate.events.external import ExternalEventProcessor
 from frigate.events.maintainer import EventProcessor
@@ -44,7 +46,8 @@ from frigate.record.record import manage_recordings
 from frigate.stats import StatsEmitter, stats_init
 from frigate.storage import StorageMaintainer
 from frigate.timeline import TimelineProcessor
-from frigate.types import CameraMetricsTypes, RecordMetricsTypes
+from frigate.types import CameraMetricsTypes, FeatureMetricsTypes
+from frigate.util import LimitedQueue as LQueue
 from frigate.version import VERSION
 from frigate.video import capture_camera, track_camera
 from frigate.watchdog import FrigateWatchdog
@@ -55,14 +58,14 @@ logger = logging.getLogger(__name__)
 class FrigateApp:
     def __init__(self) -> None:
         self.stop_event: MpEvent = mp.Event()
-        self.detection_queue: Queue = mp.Queue()
+        self.detection_queue: Queue = ff.Queue()
         self.detectors: dict[str, ObjectDetectProcess] = {}
         self.detection_out_events: dict[str, MpEvent] = {}
         self.detection_shms: list[mp.shared_memory.SharedMemory] = []
-        self.log_queue: Queue = mp.Queue()
+        self.log_queue: Queue = ff.Queue()
         self.plus_api = PlusApi()
         self.camera_metrics: dict[str, CameraMetricsTypes] = {}
-        self.record_metrics: dict[str, RecordMetricsTypes] = {}
+        self.feature_metrics: dict[str, FeatureMetricsTypes] = {}
         self.processes: dict[str, int] = {}
 
     def set_environment_vars(self) -> None:
@@ -104,37 +107,74 @@ class FrigateApp:
         user_config = FrigateConfig.parse_file(config_file)
         self.config = user_config.runtime_config(self.plus_api)
 
-        for camera_name in self.config.cameras.keys():
+        for camera_name, camera_config in self.config.cameras.items():
             # create camera_metrics
             self.camera_metrics[camera_name] = {
-                "camera_fps": mp.Value("d", 0.0),
+                "camera_fps": mp.Value("d", 0.0),  # type: ignore[typeddict-item]
-                "skipped_fps": mp.Value("d", 0.0),
+                # issue https://github.com/python/typeshed/issues/8799
-                "process_fps": mp.Value("d", 0.0),
+                # from mypy 0.981 onwards
-                "detection_enabled": mp.Value(
+                "skipped_fps": mp.Value("d", 0.0),  # type: ignore[typeddict-item]
-                    "i", self.config.cameras[camera_name].detect.enabled
+                # issue https://github.com/python/typeshed/issues/8799
+                # from mypy 0.981 onwards
+                "process_fps": mp.Value("d", 0.0),  # type: ignore[typeddict-item]
+                # issue https://github.com/python/typeshed/issues/8799
+                # from mypy 0.981 onwards
+                "detection_enabled": mp.Value(  # type: ignore[typeddict-item]
+                    # issue https://github.com/python/typeshed/issues/8799
+                    # from mypy 0.981 onwards
+                    "i",
+                    self.config.cameras[camera_name].detect.enabled,
                 ),
-                "motion_enabled": mp.Value("i", True),
+                "motion_enabled": mp.Value("i", True),  # type: ignore[typeddict-item]
-                "improve_contrast_enabled": mp.Value(
+                # issue https://github.com/python/typeshed/issues/8799
-                    "i", self.config.cameras[camera_name].motion.improve_contrast
+                # from mypy 0.981 onwards
+                "improve_contrast_enabled": mp.Value(  # type: ignore[typeddict-item]
+                    # issue https://github.com/python/typeshed/issues/8799
+                    # from mypy 0.981 onwards
+                    "i",
+                    self.config.cameras[camera_name].motion.improve_contrast,
                 ),
-                "motion_threshold": mp.Value(
+                "motion_threshold": mp.Value(  # type: ignore[typeddict-item]
-                    "i", self.config.cameras[camera_name].motion.threshold
+                    # issue https://github.com/python/typeshed/issues/8799
+                    # from mypy 0.981 onwards
+                    "i",
+                    self.config.cameras[camera_name].motion.threshold,
                 ),
-                "motion_contour_area": mp.Value(
+                "motion_contour_area": mp.Value(  # type: ignore[typeddict-item]
-                    "i", self.config.cameras[camera_name].motion.contour_area
+                    # issue https://github.com/python/typeshed/issues/8799
+                    # from mypy 0.981 onwards
+                    "i",
+                    self.config.cameras[camera_name].motion.contour_area,
                 ),
-                "detection_fps": mp.Value("d", 0.0),
+                "detection_fps": mp.Value("d", 0.0),  # type: ignore[typeddict-item]
-                "detection_frame": mp.Value("d", 0.0),
+                # issue https://github.com/python/typeshed/issues/8799
-                "read_start": mp.Value("d", 0.0),
+                # from mypy 0.981 onwards
-                "ffmpeg_pid": mp.Value("i", 0),
+                "detection_frame": mp.Value("d", 0.0),  # type: ignore[typeddict-item]
-                "frame_queue": mp.Queue(maxsize=2),
+                # issue https://github.com/python/typeshed/issues/8799
+                # from mypy 0.981 onwards
+                "read_start": mp.Value("d", 0.0),  # type: ignore[typeddict-item]
+                # issue https://github.com/python/typeshed/issues/8799
+                # from mypy 0.981 onwards
+                "ffmpeg_pid": mp.Value("i", 0),  # type: ignore[typeddict-item]
+                # issue https://github.com/python/typeshed/issues/8799
+                # from mypy 0.981 onwards
+                "frame_queue": LQueue(maxsize=2),
                 "capture_process": None,
                 "process": None,
             }
-            self.record_metrics[camera_name] = {
+            self.feature_metrics[camera_name] = {
-                "record_enabled": mp.Value(
+                "audio_enabled": mp.Value(  # type: ignore[typeddict-item]
-                    "i", self.config.cameras[camera_name].record.enabled
+                    # issue https://github.com/python/typeshed/issues/8799
-                )
+                    # from mypy 0.981 onwards
+                    "i",
+                    self.config.cameras[camera_name].audio.enabled,
+                ),
+                "record_enabled": mp.Value(  # type: ignore[typeddict-item]
+                    # issue https://github.com/python/typeshed/issues/8799
+                    # from mypy 0.981 onwards
+                    "i",
+                    self.config.cameras[camera_name].record.enabled,
+                ),
             }
 
     def set_log_levels(self) -> None:
@@ -150,22 +190,22 @@ class FrigateApp:
 
     def init_queues(self) -> None:
         # Queues for clip processing
-        self.event_queue: Queue = mp.Queue()
+        self.event_queue: Queue = ff.Queue()
-        self.event_processed_queue: Queue = mp.Queue()
+        self.event_processed_queue: Queue = ff.Queue()
-        self.video_output_queue: Queue = mp.Queue(
+        self.video_output_queue: Queue = LQueue(
             maxsize=len(self.config.cameras.keys()) * 2
         )
 
         # Queue for cameras to push tracked objects to
-        self.detected_frames_queue: Queue = mp.Queue(
+        self.detected_frames_queue: Queue = LQueue(
             maxsize=len(self.config.cameras.keys()) * 2
         )
 
         # Queue for recordings info
-        self.recordings_info_queue: Queue = mp.Queue()
+        self.recordings_info_queue: Queue = ff.Queue()
 
         # Queue for timeline events
-        self.timeline_queue: Queue = mp.Queue()
+        self.timeline_queue: Queue = ff.Queue()
 
     def init_database(self) -> None:
         def vacuum_db(db: SqliteExtDatabase) -> None:
@@ -222,7 +262,7 @@ class FrigateApp:
         recording_process = mp.Process(
             target=manage_recordings,
             name="recording_manager",
-            args=(self.config, self.recordings_info_queue, self.record_metrics),
+            args=(self.config, self.recordings_info_queue, self.feature_metrics),
         )
         recording_process.daemon = True
         self.recording_process = recording_process
@@ -281,7 +321,7 @@ class FrigateApp:
             self.config,
             self.onvif_controller,
             self.camera_metrics,
-            self.record_metrics,
+            self.feature_metrics,
             comms,
         )
 
@@ -390,6 +430,18 @@ class FrigateApp:
             capture_process.start()
             logger.info(f"Capture process started for {name}: {capture_process.pid}")
 
+    def start_audio_processors(self) -> None:
+        if len([c for c in self.config.cameras.values() if c.audio.enabled]) > 0:
+            audio_process = mp.Process(
+                target=listen_to_audio,
+                name="audio_capture",
+                args=(self.config, self.feature_metrics),
+            )
+            audio_process.daemon = True
+            audio_process.start()
+            self.processes["audioDetector"] = audio_process.pid or 0
+            logger.info(f"Audio process started: {audio_process.pid}")
+
     def start_timeline_processor(self) -> None:
         self.timeline_processor = TimelineProcessor(
             self.config, self.timeline_queue, self.stop_event
@@ -486,6 +538,7 @@ class FrigateApp:
         self.start_detected_frames_processor()
         self.start_camera_processors()
         self.start_camera_capture_processes()
+        self.start_audio_processors()
         self.start_storage_maintainer()
         self.init_stats()
         self.init_external_event_processor()

frigate/comms/dispatcher.py

@@ -6,7 +6,7 @@ from typing import Any, Callable
 
 from frigate.config import FrigateConfig
 from frigate.ptz import OnvifCommandEnum, OnvifController
-from frigate.types import CameraMetricsTypes, RecordMetricsTypes
+from frigate.types import CameraMetricsTypes, FeatureMetricsTypes
 from frigate.util import restart_frigate
 
 logger = logging.getLogger(__name__)
@@ -39,19 +39,20 @@ class Dispatcher:
         config: FrigateConfig,
         onvif: OnvifController,
         camera_metrics: dict[str, CameraMetricsTypes],
-        record_metrics: dict[str, RecordMetricsTypes],
+        feature_metrics: dict[str, FeatureMetricsTypes],
         communicators: list[Communicator],
     ) -> None:
         self.config = config
         self.onvif = onvif
         self.camera_metrics = camera_metrics
-        self.record_metrics = record_metrics
+        self.feature_metrics = feature_metrics
         self.comms = communicators
 
         for comm in self.comms:
             comm.subscribe(self._receive)
 
         self._camera_settings_handlers: dict[str, Callable] = {
+            "audio": self._on_audio_command,
             "detect": self._on_detect_command,
             "improve_contrast": self._on_motion_improve_contrast_command,
             "motion": self._on_motion_command,
@@ -186,6 +187,29 @@ class Dispatcher:
         motion_settings.threshold = payload  # type: ignore[union-attr]
         self.publish(f"{camera_name}/motion_threshold/state", payload, retain=True)
 
+    def _on_audio_command(self, camera_name: str, payload: str) -> None:
+        """Callback for audio topic."""
+        audio_settings = self.config.cameras[camera_name].audio
+
+        if payload == "ON":
+            if not self.config.cameras[camera_name].audio.enabled_in_config:
+                logger.error(
+                    "Audio detection must be enabled in the config to be turned on via MQTT."
+                )
+                return
+
+            if not audio_settings.enabled:
+                logger.info(f"Turning on audio detection for {camera_name}")
+                audio_settings.enabled = True
+                self.feature_metrics[camera_name]["audio_enabled"].value = True
+        elif payload == "OFF":
+            if self.feature_metrics[camera_name]["audio_enabled"].value:
+                logger.info(f"Turning off audio detection for {camera_name}")
+                audio_settings.enabled = False
+                self.feature_metrics[camera_name]["audio_enabled"].value = False
+
+        self.publish(f"{camera_name}/audio/state", payload, retain=True)
+
     def _on_recordings_command(self, camera_name: str, payload: str) -> None:
         """Callback for recordings topic."""
         record_settings = self.config.cameras[camera_name].record
@@ -200,12 +224,12 @@ class Dispatcher:
             if not record_settings.enabled:
                 logger.info(f"Turning on recordings for {camera_name}")
                 record_settings.enabled = True
-                self.record_metrics[camera_name]["record_enabled"].value = True
+                self.feature_metrics[camera_name]["record_enabled"].value = True
         elif payload == "OFF":
-            if self.record_metrics[camera_name]["record_enabled"].value:
+            if self.feature_metrics[camera_name]["record_enabled"].value:
                 logger.info(f"Turning off recordings for {camera_name}")
                 record_settings.enabled = False
-                self.record_metrics[camera_name]["record_enabled"].value = False
+                self.feature_metrics[camera_name]["record_enabled"].value = False
 
         self.publish(f"{camera_name}/recordings/state", payload, retain=True)
 
@@ -229,7 +253,7 @@ class Dispatcher:
         try:
             if "preset" in payload.lower():
                 command = OnvifCommandEnum.preset
-                param = payload.lower().split("-")[1]
+                param = payload.lower()[payload.index("_") + 1 :]
             else:
                 command = OnvifCommandEnum[payload.lower()]
                 param = ""

frigate/comms/mqtt.py

@@ -41,7 +41,7 @@ class MqttClient(Communicator):  # type: ignore[misc]
         for camera_name, camera in self.config.cameras.items():
             self.publish(
                 f"{camera_name}/recordings/state",
-                "ON" if camera.record.enabled else "OFF",
+                "ON" if camera.record.enabled_in_config else "OFF",
                 retain=True,
             )
             self.publish(
@@ -49,6 +49,11 @@ class MqttClient(Communicator):  # type: ignore[misc]
                 "ON" if camera.snapshots.enabled else "OFF",
                 retain=True,
             )
+            self.publish(
+                f"{camera_name}/audio/state",
+                "ON" if camera.audio.enabled_in_config else "OFF",
+                retain=True,
+            )
             self.publish(
                 f"{camera_name}/detect/state",
                 "ON" if camera.detect.enabled else "OFF",
@@ -144,6 +149,7 @@ class MqttClient(Communicator):  # type: ignore[misc]
             "recordings",
             "snapshots",
             "detect",
+            "audio",
             "motion",
             "improve_contrast",
             "motion_threshold",

frigate/config.py

@@ -40,6 +40,7 @@ DEFAULT_TIME_FORMAT = "%m/%d/%Y %H:%M:%S"
 FRIGATE_ENV_VARS = {k: v for k, v in os.environ.items() if k.startswith("FRIGATE_")}
 
 DEFAULT_TRACKED_OBJECTS = ["person"]
+DEFAULT_LISTEN_AUDIO = ["bark", "speech", "yell", "scream"]
 DEFAULT_DETECTORS = {"cpu": {"type": "cpu"}}
 
 
@@ -187,7 +188,7 @@ class RecordConfig(FrigateBaseModel):
 
 class MotionConfig(FrigateBaseModel):
     threshold: int = Field(
-        default=20,
+        default=30,
         title="Motion detection threshold (1-255).",
         ge=1,
         le=255,
@@ -387,6 +388,19 @@ class ObjectConfig(FrigateBaseModel):
     mask: Union[str, List[str]] = Field(default="", title="Object mask.")
 
 
+class AudioConfig(FrigateBaseModel):
+    enabled: bool = Field(default=False, title="Enable audio events.")
+    max_not_heard: int = Field(
+        default=30, title="Seconds of not hearing the type of audio to end the event."
+    )
+    listen: List[str] = Field(
+        default=DEFAULT_LISTEN_AUDIO, title="Audio to listen for."
+    )
+    enabled_in_config: Optional[bool] = Field(
+        title="Keep track of original state of audio detection."
+    )
+
+
 class BirdseyeModeEnum(str, Enum):
     objects = "objects"
     motion = "motion"
@@ -463,9 +477,14 @@ class FfmpegConfig(FrigateBaseModel):
         default_factory=FfmpegOutputArgsConfig,
         title="FFmpeg output arguments per role.",
     )
+    retry_interval: float = Field(
+        default=10.0,
+        title="Time in seconds to wait before FFmpeg retries connecting to the camera.",
+    )
 
 
 class CameraRoleEnum(str, Enum):
+    audio = "audio"
     record = "record"
     rtmp = "rtmp"
     detect = "detect"
@@ -627,6 +646,9 @@ class CameraConfig(FrigateBaseModel):
     objects: ObjectConfig = Field(
         default_factory=ObjectConfig, title="Object configuration."
     )
+    audio: AudioConfig = Field(
+        default_factory=AudioConfig, title="Audio events configuration."
+    )
     motion: Optional[MotionConfig] = Field(title="Motion detection configuration.")
     detect: DetectConfig = Field(
         default_factory=DetectConfig, title="Object detection configuration."
@@ -657,12 +679,16 @@ class CameraConfig(FrigateBaseModel):
         # add roles to the input if there is only one
         if len(config["ffmpeg"]["inputs"]) == 1:
             has_rtmp = "rtmp" in config["ffmpeg"]["inputs"][0].get("roles", [])
+            has_audio = "audio" in config["ffmpeg"]["inputs"][0].get("roles", [])
 
             config["ffmpeg"]["inputs"][0]["roles"] = [
                 "record",
                 "detect",
             ]
 
+            if has_audio:
+                config["ffmpeg"]["inputs"][0]["roles"].append("audio")
+
             if has_rtmp:
                 config["ffmpeg"]["inputs"][0]["roles"].append("rtmp")
 
@@ -795,6 +821,11 @@ def verify_config_roles(camera_config: CameraConfig) -> None:
             f"Camera {camera_config.name} has rtmp enabled, but rtmp is not assigned to an input."
         )
 
+    if camera_config.audio.enabled and "audio" not in assigned_roles:
+        raise ValueError(
+            f"Camera {camera_config.name} has audio events enabled, but audio is not assigned to an input."
+        )
+
 
 def verify_valid_live_stream_name(
     frigate_config: FrigateConfig, camera_config: CameraConfig
@@ -907,6 +938,9 @@ class FrigateConfig(FrigateBaseModel):
     objects: ObjectConfig = Field(
         default_factory=ObjectConfig, title="Global object configuration."
     )
+    audio: AudioConfig = Field(
+        default_factory=AudioConfig, title="Global Audio events configuration."
+    )
     motion: Optional[MotionConfig] = Field(
         title="Global motion detection configuration."
     )
@@ -931,6 +965,7 @@ class FrigateConfig(FrigateBaseModel):
         # Global config to propagate down to camera level
         global_config = config.dict(
             include={
+                "audio": ...,
                 "birdseye": ...,
                 "record": ...,
                 "snapshots": ...,
@@ -976,8 +1011,9 @@ class FrigateConfig(FrigateBaseModel):
                 camera_config.onvif.password = camera_config.onvif.password.format(
                     **FRIGATE_ENV_VARS
                 )
-            # set config recording value
+            # set config pre-value
             camera_config.record.enabled_in_config = camera_config.record.enabled
+            camera_config.audio.enabled_in_config = camera_config.audio.enabled
 
             # Add default filters
             object_keys = camera_config.objects.track

frigate/const.py

@@ -8,9 +8,26 @@ EXPORT_DIR = f"{BASE_DIR}/exports"
 BIRDSEYE_PIPE = "/tmp/cache/birdseye"
 CACHE_DIR = "/tmp/cache"
 YAML_EXT = (".yaml", ".yml")
+FRIGATE_LOCALHOST = "http://127.0.0.1:5000"
 PLUS_ENV_VAR = "PLUS_API_KEY"
 PLUS_API_HOST = "https://api.frigate.video"
-BTBN_PATH = "/usr/lib/btbn-ffmpeg"
+
+# Attributes
+
+ATTRIBUTE_LABEL_MAP = {
+    "person": ["face", "amazon"],
+    "car": ["ups", "fedex", "amazon", "license_plate"],
+}
+ALL_ATTRIBUTE_LABELS = [
+    item for sublist in ATTRIBUTE_LABEL_MAP.values() for item in sublist
+]
+
+# Audio Consts
+
+AUDIO_DURATION = 0.975
+AUDIO_FORMAT = "s16le"
+AUDIO_MAX_BIT_RANGE = 32768.0
+AUDIO_SAMPLE_RATE = 16000
 
 # Regex Consts
 
@@ -28,5 +45,4 @@ DRIVER_INTEL_iHD = "iHD"
 # Record Values
 
 MAX_SEGMENT_DURATION = 600
-SECONDS_IN_DAY = 60 * 60 * 24
 MAX_PLAYLIST_SECONDS = 7200  # support 2 hour segments for a single playlist to account for cameras with inconsistent segment times

frigate/events/audio.py

@@ -0,0 +1,255 @@
+"""Handle creating audio events."""
+
+import datetime
+import logging
+import multiprocessing as mp
+import os
+import signal
+import threading
+from types import FrameType
+from typing import Optional
+
+import numpy as np
+import requests
+from setproctitle import setproctitle
+
+from frigate.config import CameraConfig, FrigateConfig
+from frigate.const import (
+    AUDIO_DURATION,
+    AUDIO_FORMAT,
+    AUDIO_MAX_BIT_RANGE,
+    AUDIO_SAMPLE_RATE,
+    CACHE_DIR,
+    FRIGATE_LOCALHOST,
+)
+from frigate.ffmpeg_presets import parse_preset_input
+from frigate.log import LogPipe
+from frigate.object_detection import load_labels
+from frigate.types import FeatureMetricsTypes
+from frigate.util import get_ffmpeg_arg_list, listen
+from frigate.video import start_or_restart_ffmpeg, stop_ffmpeg
+
+try:
+    from tflite_runtime.interpreter import Interpreter
+except ModuleNotFoundError:
+    from tensorflow.lite.python.interpreter import Interpreter
+
+logger = logging.getLogger(__name__)
+
+
+def get_ffmpeg_command(input_args: list[str], input_path: str, pipe: str) -> list[str]:
+    return get_ffmpeg_arg_list(
+        f"ffmpeg {{}} -i {{}} -f {AUDIO_FORMAT} -ar {AUDIO_SAMPLE_RATE} -ac 1 -y {{}}".format(
+            " ".join(input_args),
+            input_path,
+            pipe,
+        )
+    )
+
+
+def listen_to_audio(
+    config: FrigateConfig,
+    process_info: dict[str, FeatureMetricsTypes],
+) -> None:
+    stop_event = mp.Event()
+    audio_threads: list[threading.Thread] = []
+
+    def exit_process() -> None:
+        for thread in audio_threads:
+            thread.join()
+
+        logger.info("Exiting audio detector...")
+
+    def receiveSignal(signalNumber: int, frame: Optional[FrameType]) -> None:
+        stop_event.set()
+        exit_process()
+
+    signal.signal(signal.SIGTERM, receiveSignal)
+    signal.signal(signal.SIGINT, receiveSignal)
+
+    threading.current_thread().name = "process:audio_manager"
+    setproctitle("frigate.audio_manager")
+    listen()
+
+    for camera in config.cameras.values():
+        if camera.enabled and camera.audio.enabled_in_config:
+            audio = AudioEventMaintainer(camera, process_info, stop_event)
+            audio_threads.append(audio)
+            audio.start()
+
+
+class AudioTfl:
+    def __init__(self, stop_event: mp.Event):
+        self.stop_event = stop_event
+        self.labels = load_labels("/audio-labelmap.txt")
+        self.interpreter = Interpreter(
+            model_path="/cpu_audio_model.tflite",
+            num_threads=2,
+        )
+
+        self.interpreter.allocate_tensors()
+
+        self.tensor_input_details = self.interpreter.get_input_details()
+        self.tensor_output_details = self.interpreter.get_output_details()
+
+    def _detect_raw(self, tensor_input):
+        self.interpreter.set_tensor(self.tensor_input_details[0]["index"], tensor_input)
+        self.interpreter.invoke()
+        detections = np.zeros((20, 6), np.float32)
+
+        res = self.interpreter.get_tensor(self.tensor_output_details[0]["index"])[0]
+        non_zero_indices = res > 0
+        class_ids = np.argpartition(-res, 20)[:20]
+        class_ids = class_ids[np.argsort(-res[class_ids])]
+        class_ids = class_ids[non_zero_indices[class_ids]]
+        scores = res[class_ids]
+        boxes = np.full((scores.shape[0], 4), -1, np.float32)
+        count = len(scores)
+
+        for i in range(count):
+            if scores[i] < 0.4 or i == 20:
+                break
+            detections[i] = [
+                class_ids[i],
+                float(scores[i]),
+                boxes[i][0],
+                boxes[i][1],
+                boxes[i][2],
+                boxes[i][3],
+            ]
+
+        return detections
+
+    def detect(self, tensor_input, threshold=0.8):
+        detections = []
+
+        if self.stop_event.is_set():
+            return detections
+
+        raw_detections = self._detect_raw(tensor_input)
+
+        for d in raw_detections:
+            if d[1] < threshold:
+                break
+            detections.append(
+                (self.labels[int(d[0])], float(d[1]), (d[2], d[3], d[4], d[5]))
+            )
+        return detections
+
+
+class AudioEventMaintainer(threading.Thread):
+    def __init__(
+        self,
+        camera: CameraConfig,
+        feature_metrics: dict[str, FeatureMetricsTypes],
+        stop_event: mp.Event,
+    ) -> None:
+        threading.Thread.__init__(self)
+        self.name = f"{camera.name}_audio_event_processor"
+        self.config = camera
+        self.feature_metrics = feature_metrics
+        self.detections: dict[dict[str, any]] = feature_metrics
+        self.stop_event = stop_event
+        self.detector = AudioTfl(stop_event)
+        self.shape = (int(round(AUDIO_DURATION * AUDIO_SAMPLE_RATE)),)
+        self.chunk_size = int(round(AUDIO_DURATION * AUDIO_SAMPLE_RATE * 2))
+        self.pipe = f"{CACHE_DIR}/{self.config.name}-audio"
+        self.ffmpeg_cmd = get_ffmpeg_command(
+            get_ffmpeg_arg_list(self.config.ffmpeg.global_args)
+            + parse_preset_input("preset-rtsp-audio-only", 1),
+            [i.path for i in self.config.ffmpeg.inputs if "audio" in i.roles][0],
+            self.pipe,
+        )
+        self.pipe_file = None
+        self.logpipe = LogPipe(f"ffmpeg.{self.config.name}.audio")
+        self.audio_listener = None
+
+    def detect_audio(self, audio) -> None:
+        if not self.feature_metrics[self.config.name]["audio_enabled"].value:
+            return
+
+        waveform = (audio / AUDIO_MAX_BIT_RANGE).astype(np.float32)
+        model_detections = self.detector.detect(waveform)
+
+        for label, score, _ in model_detections:
+            if label not in self.config.audio.listen:
+                continue
+
+            self.handle_detection(label, score)
+
+        self.expire_detections()
+
+    def handle_detection(self, label: str, score: float) -> None:
+        if self.detections.get(label):
+            self.detections[label][
+                "last_detection"
+            ] = datetime.datetime.now().timestamp()
+        else:
+            resp = requests.post(
+                f"{FRIGATE_LOCALHOST}/api/events/{self.config.name}/{label}/create",
+                json={"duration": None, "source_type": "audio"},
+            )
+
+            if resp.status_code == 200:
+                event_id = resp.json()[0]["event_id"]
+                self.detections[label] = {
+                    "id": event_id,
+                    "label": label,
+                    "last_detection": datetime.datetime.now().timestamp(),
+                }
+
+    def expire_detections(self) -> None:
+        now = datetime.datetime.now().timestamp()
+
+        for detection in self.detections.values():
+            if not detection:
+                continue
+
+            if (
+                now - detection.get("last_detection", now)
+                > self.config.audio.max_not_heard
+            ):
+                resp = requests.put(
+                    f"{FRIGATE_LOCALHOST}/api/events/{detection['id']}/end",
+                    json={
+                        "end_time": detection["last_detection"]
+                        + self.config.record.events.post_capture
+                    },
+                )
+                if resp.status_code == 200:
+                    self.detections[detection["label"]] = None
+                else:
+                    logger.warn(
+                        f"Failed to end audio event {detection['id']} with status code {resp.status_code}"
+                    )
+
+    def restart_audio_pipe(self) -> None:
+        try:
+            os.mkfifo(self.pipe)
+        except FileExistsError:
+            pass
+
+        self.audio_listener = start_or_restart_ffmpeg(
+            self.ffmpeg_cmd, logger, self.logpipe, None, self.audio_listener
+        )
+
+    def read_audio(self) -> None:
+        if self.pipe_file is None:
+            self.pipe_file = open(self.pipe, "rb")
+
+        try:
+            audio = np.frombuffer(self.pipe_file.read(self.chunk_size), dtype=np.int16)
+            self.detect_audio(audio)
+        except BrokenPipeError:
+            self.logpipe.dump()
+            self.restart_audio_pipe()
+
+    def run(self) -> None:
+        self.restart_audio_pipe()
+
+        while not self.stop_event.is_set():
+            self.read_audio()
+
+        self.pipe_file.close()
+        stop_ffmpeg(self.audio_listener, logger)
+        self.logpipe.close()

frigate/events/external.py

@@ -6,10 +6,10 @@ import logging
 import os
 import random
 import string
-from multiprocessing.queues import Queue
 from typing import Optional
 
 import cv2
+from faster_fifo import Queue
 
 from frigate.config import CameraConfig, FrigateConfig
 from frigate.const import CLIPS_DIR
@@ -29,6 +29,7 @@ class ExternalEventProcessor:
         self,
         camera: str,
         label: str,
+        source_type: str,
         sub_label: Optional[str],
         duration: Optional[int],
         include_recording: bool,
@@ -56,22 +57,26 @@ class ExternalEventProcessor:
                     "label": label,
                     "sub_label": sub_label,
                     "camera": camera,
-                    "start_time": now,
+                    "start_time": now - camera_config.record.events.pre_capture,
-                    "end_time": now + duration if duration is not None else None,
+                    "end_time": now
+                    + duration
+                    + camera_config.record.events.post_capture
+                    if duration is not None
+                    else None,
                     "thumbnail": thumbnail,
                     "has_clip": camera_config.record.enabled and include_recording,
                     "has_snapshot": True,
+                    "type": source_type,
                 },
             )
         )
 
         return event_id
 
-    def finish_manual_event(self, event_id: str) -> None:
+    def finish_manual_event(self, event_id: str, end_time: float) -> None:
         """Finish external event with indeterminate duration."""
-        now = datetime.datetime.now().timestamp()
         self.queue.put(
-            (EventTypeEnum.api, "end", None, {"id": event_id, "end_time": now})
+            (EventTypeEnum.api, "end", None, {"id": event_id, "end_time": end_time})
         )
 
     def _write_images(

frigate/events/maintainer.py

@@ -3,10 +3,11 @@ import logging
 import queue
 import threading
 from enum import Enum
-from multiprocessing.queues import Queue
 from multiprocessing.synchronize import Event as MpEvent
 from typing import Dict
 
+from faster_fifo import Queue
+
 from frigate.config import EventsConfig, FrigateConfig
 from frigate.models import Event
 from frigate.types import CameraMetricsTypes
@@ -17,7 +18,6 @@ logger = logging.getLogger(__name__)
 
 class EventTypeEnum(str, Enum):
     api = "api"
-    # audio = "audio"
     tracked_object = "tracked_object"
 
 
@@ -72,8 +72,11 @@ class EventProcessor(threading.Thread):
             except queue.Empty:
                 continue
 
-            logger.debug(f"Event received: {event_type} {camera} {event_data['id']}")
+            logger.debug(
+                f"Event received: {source_type} {event_type} {camera} {event_data['id']}"
+            )
 
+            if source_type == EventTypeEnum.tracked_object:
                 self.timeline_queue.put(
                     (
                         camera,
@@ -84,7 +87,6 @@ class EventProcessor(threading.Thread):
                     )
                 )
 
-            if source_type == EventTypeEnum.tracked_object:
                 if event_type == "start":
                     self.events_in_process[event_data["id"]] = event_data
                     continue
@@ -191,6 +193,7 @@ class EventProcessor(threading.Thread):
                     "score": score,
                     "top_score": event_data["top_score"],
                     "attributes": attributes,
+                    "type": "object",
                 },
             }
 
@@ -214,8 +217,8 @@ class EventProcessor(threading.Thread):
             del self.events_in_process[event_data["id"]]
             self.event_processed_queue.put((event_data["id"], camera))
 
-    def handle_external_detection(self, type: str, event_data: Event):
+    def handle_external_detection(self, event_type: str, event_data: Event) -> None:
-        if type == "new":
+        if event_type == "new":
             event = {
                 Event.id: event_data["id"],
                 Event.label: event_data["label"],
@@ -227,22 +230,16 @@ class EventProcessor(threading.Thread):
                 Event.has_clip: event_data["has_clip"],
                 Event.has_snapshot: event_data["has_snapshot"],
                 Event.zones: [],
-                Event.data: {},
+                Event.data: {"type": event_data["type"]},
             }
-        elif type == "end":
+            Event.insert(event).execute()
+        elif event_type == "end":
             event = {
                 Event.id: event_data["id"],
                 Event.end_time: event_data["end_time"],
             }
 
             try:
-            (
+                Event.update(event).where(Event.id == event_data["id"]).execute()
-                Event.insert(event)
-                .on_conflict(
-                    conflict_target=[Event.id],
-                    update=event,
-                )
-                .execute()
-            )
             except Exception:
                 logger.warning(f"Failed to update manual event: {event_data['id']}")

frigate/ffmpeg_presets.py

@@ -5,7 +5,6 @@ import os
 from enum import Enum
 from typing import Any
 
-from frigate.const import BTBN_PATH
 from frigate.util import vainfo_hwaccel
 from frigate.version import VERSION
 
@@ -43,7 +42,11 @@ class LibvaGpuSelector:
         return ""
 
 
-TIMEOUT_PARAM = "-timeout" if os.path.exists(BTBN_PATH) else "-stimeout"
+TIMEOUT_PARAM = (
+    "-timeout"
+    if int(os.getenv("LIBAVFORMAT_VERSION_MAJOR", "59")) >= 59
+    else "-stimeout"
+)
 
 _gpu_selector = LibvaGpuSelector()
 _user_agent_args = [
@@ -107,14 +110,14 @@ PRESETS_HW_ACCEL_DECODE = {
 }
 
 PRESETS_HW_ACCEL_SCALE = {
-    "preset-rpi-32-h264": "-r {0} -s {1}x{2}",
+    "preset-rpi-32-h264": "-r {0} -vf fps={0},scale={1}:{2}",
-    "preset-rpi-64-h264": "-r {0} -s {1}x{2}",
+    "preset-rpi-64-h264": "-r {0} -vf fps={0},scale={1}:{2}",
     "preset-vaapi": "-r {0} -vf fps={0},scale_vaapi=w={1}:h={2},hwdownload,format=yuv420p",
     "preset-intel-qsv-h264": "-r {0} -vf vpp_qsv=framerate={0}:w={1}:h={2}:format=nv12,hwdownload,format=nv12,format=yuv420p",
     "preset-intel-qsv-h265": "-r {0} -vf vpp_qsv=framerate={0}:w={1}:h={2}:format=nv12,hwdownload,format=nv12,format=yuv420p",
     "preset-nvidia-h264": "-r {0} -vf fps={0},scale_cuda=w={1}:h={2}:format=nv12,hwdownload,format=nv12,format=yuv420p",
     "preset-nvidia-h265": "-r {0} -vf fps={0},scale_cuda=w={1}:h={2}:format=nv12,hwdownload,format=nv12,format=yuv420p",
-    "default": "-r {0} -s {1}x{2}",
+    "default": "-r {0} -vf fps={0},scale={1}:{2}",
 }
 
 PRESETS_HW_ACCEL_ENCODE_BIRDSEYE = {
@@ -282,6 +285,13 @@ PRESETS_INPUT = {
         "-use_wallclock_as_timestamps",
         "1",
     ],
+    "preset-rtsp-audio-only": [
+        "-rtsp_transport",
+        "tcp",
+        TIMEOUT_PARAM,
+        "5000000",
+        "-vn",
+    ],
     "preset-rtsp-restream": _user_agent_args
     + [
         "-rtsp_transport",

frigate/http.py

@@ -410,6 +410,24 @@ def set_sub_label(id):
     )
 
 
+@bp.route("/labels")
+def get_labels():
+    camera = request.args.get("camera", type=str, default="")
+
+    try:
+        if camera:
+            events = Event.select(Event.label).where(Event.camera == camera).distinct()
+        else:
+            events = Event.select(Event.label).distinct()
+    except Exception as e:
+        return jsonify(
+            {"success": False, "message": f"Failed to get labels: {e}"}, "404"
+        )
+
+    labels = sorted([e.label for e in events])
+    return jsonify(labels)
+
+
 @bp.route("/sub_labels")
 def get_sub_labels():
     split_joined = request.args.get("split_joined", type=int)
@@ -866,6 +884,7 @@ def create_event(camera_name, label):
         event_id = current_app.external_processor.create_manual_event(
             camera_name,
             label,
+            json.get("source_type", "api"),
             json.get("sub_label", None),
             json.get("duration", 30),
             json.get("include_recording", True),
@@ -890,8 +909,11 @@ def create_event(camera_name, label):
 
 @bp.route("/events/<event_id>/end", methods=["PUT"])
 def end_event(event_id):
+    json: dict[str, any] = request.get_json(silent=True) or {}
+
     try:
-        current_app.external_processor.finish_manual_event(event_id)
+        end_time = json.get("end_time", datetime.now().timestamp())
+        current_app.external_processor.finish_manual_event(event_id, end_time)
     except Exception:
         return jsonify(
             {"success": False, "message": f"{event_id} must be set and valid."}, 404
@@ -1100,6 +1122,15 @@ def latest_frame(camera_name):
         height = int(request.args.get("h", str(frame.shape[0])))
         width = int(height * frame.shape[1] / frame.shape[0])
 
+        if frame is None:
+            return "Unable to get valid frame from {}".format(camera_name), 500
+
+        if height < 1 or width < 1:
+            return (
+                "Invalid height / width requested :: {} / {}".format(height, width),
+                400,
+            )
+
         frame = cv2.resize(frame, dsize=(width, height), interpolation=cv2.INTER_AREA)
 
         ret, jpg = cv2.imencode(

frigate/log.py

@@ -7,10 +7,10 @@ import signal
 import threading
 from collections import deque
 from logging import handlers
-from multiprocessing.queues import Queue
 from types import FrameType
 from typing import Deque, Optional
 
+from faster_fifo import Queue
 from setproctitle import setproctitle
 
 from frigate.util import clean_camera_user_pass

frigate/motion/improved_motion.py

@@ -1,6 +1,7 @@
 import cv2
 import imutils
 import numpy as np
+from scipy.ndimage import gaussian_filter
 
 from frigate.config import MotionConfig
 from frigate.motion import MotionDetector
@@ -15,9 +16,10 @@ class ImprovedMotionDetector(MotionDetector):
         improve_contrast,
         threshold,
         contour_area,
-        clipLimit=2.0,
-        tileGridSize=(2, 2),
         name="improved",
+        blur_radius=1,
+        interpolation=cv2.INTER_NEAREST,
+        contrast_frame_history=50,
     ):
         self.name = name
         self.config = config
@@ -28,13 +30,12 @@ class ImprovedMotionDetector(MotionDetector):
             config.frame_height * frame_shape[1] // frame_shape[0],
         )
         self.avg_frame = np.zeros(self.motion_frame_size, np.float32)
-        self.avg_delta = np.zeros(self.motion_frame_size, np.float32)
         self.motion_frame_count = 0
         self.frame_counter = 0
         resized_mask = cv2.resize(
             config.mask,
             dsize=(self.motion_frame_size[1], self.motion_frame_size[0]),
-            interpolation=cv2.INTER_LINEAR,
+            interpolation=cv2.INTER_AREA,
         )
         self.mask = np.where(resized_mask == [0])
         self.save_images = False
@@ -42,7 +43,11 @@ class ImprovedMotionDetector(MotionDetector):
         self.improve_contrast = improve_contrast
         self.threshold = threshold
         self.contour_area = contour_area
-        self.clahe = cv2.createCLAHE(clipLimit=clipLimit, tileGridSize=tileGridSize)
+        self.blur_radius = blur_radius
+        self.interpolation = interpolation
+        self.contrast_values = np.zeros((contrast_frame_history, 2), np.uint8)
+        self.contrast_values[:, 1:2] = 255
+        self.contrast_values_index = 0
 
     def detect(self, frame):
         motion_boxes = []
@@ -53,27 +58,44 @@ class ImprovedMotionDetector(MotionDetector):
         resized_frame = cv2.resize(
             gray,
             dsize=(self.motion_frame_size[1], self.motion_frame_size[0]),
-            interpolation=cv2.INTER_LINEAR,
+            interpolation=self.interpolation,
         )
 
         if self.save_images:
             resized_saved = resized_frame.copy()
 
-        resized_frame = cv2.GaussianBlur(resized_frame, (3, 3), cv2.BORDER_DEFAULT)
-
-        if self.save_images:
-            blurred_saved = resized_frame.copy()
-
         # Improve contrast
         if self.improve_contrast.value:
-            resized_frame = self.clahe.apply(resized_frame)
+            # TODO tracking moving average of min/max to avoid sudden contrast changes
+            minval = np.percentile(resized_frame, 4).astype(np.uint8)
+            maxval = np.percentile(resized_frame, 96).astype(np.uint8)
+            # skip contrast calcs if the image is a single color
+            if minval < maxval:
+                # keep track of the last 50 contrast values
+                self.contrast_values[self.contrast_values_index] = [minval, maxval]
+                self.contrast_values_index += 1
+                if self.contrast_values_index == len(self.contrast_values):
+                    self.contrast_values_index = 0
+
+                avg_min, avg_max = np.mean(self.contrast_values, axis=0)
+
+                resized_frame = np.clip(resized_frame, avg_min, avg_max)
+                resized_frame = (
+                    ((resized_frame - avg_min) / (avg_max - avg_min)) * 255
+                ).astype(np.uint8)
 
         if self.save_images:
             contrasted_saved = resized_frame.copy()
 
         # mask frame
+        # this has to come after contrast improvement
         resized_frame[self.mask] = [255]
 
+        resized_frame = gaussian_filter(resized_frame, sigma=1, radius=self.blur_radius)
+
+        if self.save_images:
+            blurred_saved = resized_frame.copy()
+
         if self.save_images or self.calibrating:
             self.frame_counter += 1
         # compare to average
@@ -134,8 +156,8 @@ class ImprovedMotionDetector(MotionDetector):
                 )
             frames = [
                 cv2.cvtColor(resized_saved, cv2.COLOR_GRAY2BGR),
-                cv2.cvtColor(blurred_saved, cv2.COLOR_GRAY2BGR),
                 cv2.cvtColor(contrasted_saved, cv2.COLOR_GRAY2BGR),
+                cv2.cvtColor(blurred_saved, cv2.COLOR_GRAY2BGR),
                 cv2.cvtColor(frameDelta, cv2.COLOR_GRAY2BGR),
                 cv2.cvtColor(thresh, cv2.COLOR_GRAY2BGR),
                 thresh_dilated,

frigate/object_detection.py

@@ -7,6 +7,7 @@ import signal
 import threading
 from abc import ABC, abstractmethod
 
+import faster_fifo as ff
 import numpy as np
 from setproctitle import setproctitle
 
@@ -72,7 +73,7 @@ class LocalObjectDetector(ObjectDetector):
 
 def run_detector(
     name: str,
-    detection_queue: mp.Queue,
+    detection_queue: ff.Queue,
     out_events: dict[str, mp.Event],
     avg_speed,
     start,

frigate/object_processing.py

@@ -188,15 +188,14 @@ class TrackedObject:
             zone_score = self.zone_presence.get(name, 0)
             # check if the object is in the zone
             if cv2.pointPolygonTest(contour, bottom_center, False) >= 0:
+                # if the object passed the filters once, dont apply again
+                if name in self.current_zones or not zone_filtered(self, zone.filters):
                     self.zone_presence[name] = zone_score + 1
 
                     # an object is only considered present in a zone if it has a zone inertia of 3+
                     if zone_score >= zone.inertia:
-                    # if the object passed the filters once, dont apply again
-                    if name in self.current_zones or not zone_filtered(
-                        self, zone.filters
-                    ):
                         current_zones.append(name)
+
                         if name not in self.entered_zones:
                             self.entered_zones.append(name)
             else:

frigate/output.py

@@ -29,6 +29,61 @@ from frigate.util import SharedMemoryFrameManager, copy_yuv_to_position, get_yuv
 logger = logging.getLogger(__name__)
 
 
+def get_standard_aspect_ratio(width, height) -> tuple[int, int]:
+    """Ensure that only standard aspect ratios are used."""
+    known_aspects = [
+        (16, 9),
+        (9, 16),
+        (32, 9),
+        (12, 9),
+        (9, 12),
+    ]  # aspects are scaled to have common relative size
+    known_aspects_ratios = list(
+        map(lambda aspect: aspect[0] / aspect[1], known_aspects)
+    )
+    closest = min(
+        known_aspects_ratios,
+        key=lambda x: abs(x - (width / height)),
+    )
+    return known_aspects[known_aspects_ratios.index(closest)]
+
+
+class Canvas:
+    def __init__(self, canvas_width: int, canvas_height: int) -> None:
+        gcd = math.gcd(canvas_width, canvas_height)
+        self.aspect = get_standard_aspect_ratio(
+            (canvas_width / gcd), (canvas_height / gcd)
+        )
+        self.width = canvas_width
+        self.height = (self.width * self.aspect[1]) / self.aspect[0]
+        self.coefficient_cache: dict[int, int] = {}
+        self.aspect_cache: dict[str, tuple[int, int]] = {}
+
+    def get_aspect(self, coefficient: int) -> tuple[int, int]:
+        return (self.aspect[0] * coefficient, self.aspect[1] * coefficient)
+
+    def get_coefficient(self, camera_count: int) -> int:
+        return self.coefficient_cache.get(camera_count, 2)
+
+    def set_coefficient(self, camera_count: int, coefficient: int) -> None:
+        self.coefficient_cache[camera_count] = coefficient
+
+    def get_camera_aspect(
+        self, cam_name: str, camera_width: int, camera_height: int
+    ) -> tuple[int, int]:
+        cached = self.aspect_cache.get(cam_name)
+
+        if cached:
+            return cached
+
+        gcd = math.gcd(camera_width, camera_height)
+        camera_aspect = get_standard_aspect_ratio(
+            camera_width / gcd, camera_height / gcd
+        )
+        self.aspect_cache[cam_name] = camera_aspect
+        return camera_aspect
+
+
 class FFMpegConverter:
     def __init__(
         self,
@@ -156,7 +211,12 @@ class BroadcastThread(threading.Thread):
 
 
 class BirdsEyeFrameManager:
-    def __init__(self, config: FrigateConfig, frame_manager: SharedMemoryFrameManager):
+    def __init__(
+        self,
+        config: FrigateConfig,
+        frame_manager: SharedMemoryFrameManager,
+        stop_event: mp.Event,
+    ):
         self.config = config
         self.mode = config.birdseye.mode
         self.frame_manager = frame_manager
@@ -165,6 +225,8 @@ class BirdsEyeFrameManager:
         self.frame_shape = (height, width)
         self.yuv_shape = (height * 3 // 2, width)
         self.frame = np.ndarray(self.yuv_shape, dtype=np.uint8)
+        self.canvas = Canvas(width, height)
+        self.stop_event = stop_event
 
         # initialize the frame as black and with the Frigate logo
         self.blank_frame = np.zeros(self.yuv_shape, np.uint8)
@@ -270,15 +332,152 @@ class BirdsEyeFrameManager:
     def update_frame(self):
         """Update to a new frame for birdseye."""
 
-        def calculate_layout(
+        # determine how many cameras are tracking objects within the last 30 seconds
-            canvas, cameras_to_add: list[str], coefficient
+        active_cameras = set(
-        ) -> tuple[any]:
+            [
+                cam
+                for cam, cam_data in self.cameras.items()
+                if cam_data["last_active_frame"] > 0
+                and cam_data["current_frame"] - cam_data["last_active_frame"] < 30
+            ]
+        )
+
+        # if there are no active cameras
+        if len(active_cameras) == 0:
+            # if the layout is already cleared
+            if len(self.camera_layout) == 0:
+                return False
+            # if the layout needs to be cleared
+            else:
+                self.camera_layout = []
+                self.active_cameras = set()
+                self.clear_frame()
+                return True
+
+        # check if we need to reset the layout because there is a different number of cameras
+        reset_layout = len(self.active_cameras) - len(active_cameras) != 0
+
+        # reset the layout if it needs to be different
+        if reset_layout:
+            logger.debug("Added new cameras, resetting layout...")
+            self.clear_frame()
+            self.active_cameras = active_cameras
+
+            # this also converts added_cameras from a set to a list since we need
+            # to pop elements in order
+            active_cameras_to_add = sorted(
+                active_cameras,
+                # sort cameras by order and by name if the order is the same
+                key=lambda active_camera: (
+                    self.config.cameras[active_camera].birdseye.order,
+                    active_camera,
+                ),
+            )
+
+            if len(active_cameras) == 1:
+                # show single camera as fullscreen
+                camera = active_cameras_to_add[0]
+                camera_dims = self.cameras[camera]["dimensions"].copy()
+                scaled_width = int(self.canvas.height * camera_dims[0] / camera_dims[1])
+                coefficient = (
+                    1
+                    if scaled_width <= self.canvas.width
+                    else self.canvas.width / scaled_width
+                )
+                self.camera_layout = [
+                    [
+                        (
+                            camera,
+                            (
+                                0,
+                                0,
+                                int(scaled_width * coefficient),
+                                int(self.canvas.height * coefficient),
+                            ),
+                        )
+                    ]
+                ]
+            else:
+                # calculate optimal layout
+                coefficient = self.canvas.get_coefficient(len(active_cameras))
+                calculating = True
+
+                # decrease scaling coefficient until height of all cameras can fit into the birdseye canvas
+                while calculating:
+                    if self.stop_event.is_set():
+                        return
+
+                    layout_candidate = self.calculate_layout(
+                        active_cameras_to_add,
+                        coefficient,
+                    )
+
+                    if not layout_candidate:
+                        if coefficient < 10:
+                            coefficient += 1
+                            continue
+                        else:
+                            logger.error("Error finding appropriate birdseye layout")
+                            return
+
+                    calculating = False
+                    self.canvas.set_coefficient(len(active_cameras), coefficient)
+
+                self.camera_layout = layout_candidate
+
+        for row in self.camera_layout:
+            for position in row:
+                self.copy_to_position(
+                    position[1], position[0], self.cameras[position[0]]["current_frame"]
+                )
+
+        return True
+
+    def calculate_layout(self, cameras_to_add: list[str], coefficient) -> tuple[any]:
         """Calculate the optimal layout for 2+ cameras."""
+
+        def map_layout(row_height: int):
+            """Map the calculated layout."""
+            candidate_layout = []
+            starting_x = 0
+            x = 0
+            max_width = 0
+            y = 0
+
+            for row in camera_layout:
+                final_row = []
+                max_width = max(max_width, x)
+                x = starting_x
+                for cameras in row:
+                    camera_dims = self.cameras[cameras[0]]["dimensions"].copy()
+                    camera_aspect = cameras[1]
+
+                    if camera_dims[1] > camera_dims[0]:
+                        scaled_height = int(row_height * 2)
+                        scaled_width = int(scaled_height * camera_aspect)
+                        starting_x = scaled_width
+                    else:
+                        scaled_height = row_height
+                        scaled_width = int(scaled_height * camera_aspect)
+
+                    # layout is too large
+                    if (
+                        x + scaled_width > self.canvas.width
+                        or y + scaled_height > self.canvas.height
+                    ):
+                        return 0, 0, None
+
+                    final_row.append((cameras[0], (x, y, scaled_width, scaled_height)))
+                    x += scaled_width
+
+                y += row_height
+                candidate_layout.append(final_row)
+
+            return max_width, y, candidate_layout
+
+        canvas_aspect_x, canvas_aspect_y = self.canvas.get_aspect(coefficient)
         camera_layout: list[list[any]] = []
         camera_layout.append([])
-            canvas_gcd = math.gcd(canvas[0], canvas[1])
-            canvas_aspect_x = (canvas[0] / canvas_gcd) * coefficient
-            canvas_aspect_y = (canvas[0] / canvas_gcd) * coefficient
         starting_x = 0
         x = starting_x
         y = 0
@@ -286,18 +485,9 @@ class BirdsEyeFrameManager:
         max_y = 0
         for camera in cameras_to_add:
             camera_dims = self.cameras[camera]["dimensions"].copy()
-                camera_gcd = math.gcd(camera_dims[0], camera_dims[1])
+            camera_aspect_x, camera_aspect_y = self.canvas.get_camera_aspect(
-                camera_aspect_x = camera_dims[0] / camera_gcd
+                camera, camera_dims[0], camera_dims[1]
-                camera_aspect_y = camera_dims[1] / camera_gcd
+            )
-
-                if round(camera_aspect_x / camera_aspect_y, 1) == 1.8:
-                    # account for slightly off 16:9 cameras
-                    camera_aspect_x = 16
-                    camera_aspect_y = 9
-                elif round(camera_aspect_x / camera_aspect_y, 1) == 1.3:
-                    # make 4:3 cameras the same relative size as 16:9
-                    camera_aspect_x = 12
-                    camera_aspect_y = 9
 
             if camera_dims[1] > camera_dims[0]:
                 portrait = True
@@ -309,10 +499,7 @@ class BirdsEyeFrameManager:
                 camera_layout[y_i].append(
                     (
                         camera,
-                            (
+                        camera_aspect_x / camera_aspect_y,
-                                camera_aspect_x,
-                                camera_aspect_y,
-                            ),
                     )
                 )
 
@@ -338,7 +525,7 @@ class BirdsEyeFrameManager:
                 camera_layout[y_i].append(
                     (
                         camera,
-                            (camera_aspect_x, camera_aspect_y),
+                        camera_aspect_x / camera_aspect_y,
                     )
                 )
                 x += camera_aspect_x
@@ -346,143 +533,24 @@ class BirdsEyeFrameManager:
         if y + max_y > canvas_aspect_y:
             return None
 
-            row_height = int(canvas_height / coefficient)
+        row_height = int(self.canvas.height / coefficient)
+        total_width, total_height, standard_candidate_layout = map_layout(row_height)
 
-            final_camera_layout = []
+        # layout can't be optimized more
-            starting_x = 0
+        if total_width / self.canvas.width >= 0.99:
-            y = 0
+            return standard_candidate_layout
 
-            for row in camera_layout:
+        scale_up_percent = min(
-                final_row = []
+            1 - (total_width / self.canvas.width),
-                x = starting_x
+            1 - (total_height / self.canvas.height),
-                for cameras in row:
-                    camera_dims = self.cameras[cameras[0]]["dimensions"].copy()
-
-                    if camera_dims[1] > camera_dims[0]:
-                        scaled_height = int(row_height * coefficient)
-                        scaled_width = int(
-                            scaled_height * camera_dims[0] / camera_dims[1]
         )
-                        starting_x = scaled_width
+        row_height = int(row_height * (1 + round(scale_up_percent, 1)))
+        _, _, scaled_layout = map_layout(row_height)
+
+        if scaled_layout:
+            return scaled_layout
         else:
-                        scaled_height = row_height
+            return standard_candidate_layout
-                        scaled_width = int(
-                            scaled_height * camera_dims[0] / camera_dims[1]
-                        )
-
-                    if (
-                        x + scaled_width > canvas_width
-                        or y + scaled_height > canvas_height
-                    ):
-                        return None
-
-                    final_row.append((cameras[0], (x, y, scaled_width, scaled_height)))
-                    x += scaled_width
-                y += row_height
-                final_camera_layout.append(final_row)
-
-            return final_camera_layout
-
-        # determine how many cameras are tracking objects within the last 30 seconds
-        active_cameras = set(
-            [
-                cam
-                for cam, cam_data in self.cameras.items()
-                if cam_data["last_active_frame"] > 0
-                and cam_data["current_frame"] - cam_data["last_active_frame"] < 30
-            ]
-        )
-
-        # if there are no active cameras
-        if len(active_cameras) == 0:
-            # if the layout is already cleared
-            if len(self.camera_layout) == 0:
-                return False
-            # if the layout needs to be cleared
-            else:
-                self.camera_layout = []
-                self.active_cameras = set()
-                self.clear_frame()
-                return True
-
-        # check if we need to reset the layout because there are new cameras to add
-        reset_layout = (
-            True if len(active_cameras.difference(self.active_cameras)) > 0 else False
-        )
-
-        # reset the layout if it needs to be different
-        if reset_layout:
-            logger.debug("Added new cameras, resetting layout...")
-            self.clear_frame()
-            self.active_cameras = active_cameras
-
-            # this also converts added_cameras from a set to a list since we need
-            # to pop elements in order
-            active_cameras_to_add = sorted(
-                active_cameras,
-                # sort cameras by order and by name if the order is the same
-                key=lambda active_camera: (
-                    self.config.cameras[active_camera].birdseye.order,
-                    active_camera,
-                ),
-            )
-
-            canvas_width = self.config.birdseye.width
-            canvas_height = self.config.birdseye.height
-
-            if len(active_cameras) == 1:
-                # show single camera as fullscreen
-                camera = active_cameras_to_add[0]
-                camera_dims = self.cameras[camera]["dimensions"].copy()
-                scaled_width = int(canvas_height * camera_dims[0] / camera_dims[1])
-                coefficient = (
-                    1 if scaled_width <= canvas_width else canvas_width / scaled_width
-                )
-                self.camera_layout = [
-                    [
-                        (
-                            camera,
-                            (
-                                0,
-                                0,
-                                int(scaled_width * coefficient),
-                                int(canvas_height * coefficient),
-                            ),
-                        )
-                    ]
-                ]
-            else:
-                # calculate optimal layout
-                coefficient = 2
-                calculating = True
-
-                # decrease scaling coefficient until height of all cameras can fit into the birdseye canvas
-                while calculating:
-                    layout_candidate = calculate_layout(
-                        (canvas_width, canvas_height),
-                        active_cameras_to_add,
-                        coefficient,
-                    )
-
-                    if not layout_candidate:
-                        if coefficient < 10:
-                            coefficient += 1
-                            continue
-                        else:
-                            logger.error("Error finding appropriate birdseye layout")
-                            return
-
-                    calculating = False
-
-                self.camera_layout = layout_candidate
-
-        for row in self.camera_layout:
-            for position in row:
-                self.copy_to_position(
-                    position[1], position[0], self.cameras[position[0]]["current_frame"]
-                )
-
-        return True
 
     def update(self, camera, object_count, motion_count, frame_time, frame) -> bool:
         # don't process if birdseye is disabled for this camera
@@ -580,7 +648,7 @@ def output_frames(config: FrigateConfig, video_output_queue):
     for t in broadcasters.values():
         t.start()
 
-    birdseye_manager = BirdsEyeFrameManager(config, frame_manager)
+    birdseye_manager = BirdsEyeFrameManager(config, frame_manager, stop_event)
 
     if config.birdseye.restream:
         birdseye_buffer = frame_manager.create(

frigate/record/cleanup.py

@@ -8,10 +8,10 @@ import threading
 from multiprocessing.synchronize import Event as MpEvent
 from pathlib import Path
 
-from peewee import DatabaseError, DoesNotExist, chunked
+from peewee import DatabaseError, chunked
 
 from frigate.config import FrigateConfig, RetainModeEnum
-from frigate.const import RECORD_DIR, SECONDS_IN_DAY
+from frigate.const import RECORD_DIR
 from frigate.models import Event, Recordings, RecordingsToDelete, Timeline
 from frigate.record.util import remove_empty_directories
 
@@ -28,7 +28,7 @@ class RecordingCleanup(threading.Thread):
         self.stop_event = stop_event
 
     def clean_tmp_clips(self) -> None:
-        # delete any clips more than 5 minutes old
+        """delete any clips in the cache that are more than 5 minutes old."""
         for p in Path("/tmp/cache").rglob("clip_*.mp4"):
             logger.debug(f"Checking tmp clip {p}.")
             if p.stat().st_mtime < (datetime.datetime.now().timestamp() - 60 * 1):
@@ -40,8 +40,8 @@ class RecordingCleanup(threading.Thread):
                 p.unlink(missing_ok=True)
 
     def expire_recordings(self) -> None:
-        logger.debug("Start expire recordings (new).")
+        """Delete recordings based on retention config."""
-
+        logger.debug("Start expire recordings.")
         logger.debug("Start deleted cameras.")
         # Handle deleted cameras
         expire_days = self.config.record.retain.days
@@ -161,59 +161,10 @@ class RecordingCleanup(threading.Thread):
             logger.debug(f"End camera: {camera}.")
 
         logger.debug("End all cameras.")
-        logger.debug("End expire recordings (new).")
+        logger.debug("End expire recordings.")
-
-    def expire_files(self) -> None:
-        logger.debug("Start expire files (legacy).")
-
-        default_expire = (
-            datetime.datetime.now().timestamp()
-            - SECONDS_IN_DAY * self.config.record.retain.days
-        )
-        delete_before = {}
-
-        for name, camera in self.config.cameras.items():
-            delete_before[name] = (
-                datetime.datetime.now().timestamp()
-                - SECONDS_IN_DAY * camera.record.retain.days
-            )
-
-        # find all the recordings older than the oldest recording in the db
-        try:
-            oldest_recording = (
-                Recordings.select().order_by(Recordings.start_time).limit(1).get()
-            )
-
-            p = Path(oldest_recording.path)
-            oldest_timestamp = p.stat().st_mtime - 1
-        except DoesNotExist:
-            oldest_timestamp = datetime.datetime.now().timestamp()
-        except FileNotFoundError:
-            logger.warning(f"Unable to find file from recordings database: {p}")
-            Recordings.delete().where(Recordings.id == oldest_recording.id).execute()
-            return
-
-        logger.debug(f"Oldest recording in the db: {oldest_timestamp}")
-
-        files_to_check = []
-
-        for root, _, files in os.walk(RECORD_DIR):
-            for file in files:
-                file_path = os.path.join(root, file)
-                if os.path.getmtime(file_path) < oldest_timestamp:
-                    files_to_check.append(file_path)
-
-        for f in files_to_check:
-            p = Path(f)
-            try:
-                if p.stat().st_mtime < delete_before.get(p.parent.name, default_expire):
-                    p.unlink(missing_ok=True)
-            except FileNotFoundError:
-                logger.warning(f"Attempted to expire missing file: {f}")
-
-        logger.debug("End expire files (legacy).")
 
     def sync_recordings(self) -> None:
+        """Check the db for stale recordings entries that don't exist in the filesystem."""
         logger.debug("Start sync recordings.")
 
         # get all recordings in the db
@@ -283,5 +234,4 @@ class RecordingCleanup(threading.Thread):
 
             if counter == 0:
                 self.expire_recordings()
-                self.expire_files()
                 remove_empty_directories(RECORD_DIR)

frigate/record/maintainer.py

@@ -3,7 +3,6 @@
 import asyncio
 import datetime
 import logging
-import multiprocessing as mp
 import os
 import queue
 import random
@@ -15,12 +14,13 @@ from multiprocessing.synchronize import Event as MpEvent
 from pathlib import Path
 from typing import Any, Tuple
 
+import faster_fifo as ff
 import psutil
 
 from frigate.config import FrigateConfig, RetainModeEnum
 from frigate.const import CACHE_DIR, MAX_SEGMENT_DURATION, RECORD_DIR
 from frigate.models import Event, Recordings
-from frigate.types import RecordMetricsTypes
+from frigate.types import FeatureMetricsTypes
 from frigate.util import area, get_video_properties
 
 logger = logging.getLogger(__name__)
@@ -30,8 +30,8 @@ class RecordingMaintainer(threading.Thread):
     def __init__(
         self,
         config: FrigateConfig,
-        recordings_info_queue: mp.Queue,
+        recordings_info_queue: ff.Queue,
-        process_info: dict[str, RecordMetricsTypes],
+        process_info: dict[str, FeatureMetricsTypes],
         stop_event: MpEvent,
     ):
         threading.Thread.__init__(self)

frigate/record/record.py

@@ -7,6 +7,7 @@ import threading
 from types import FrameType
 from typing import Optional
 
+import faster_fifo as ff
 from playhouse.sqliteq import SqliteQueueDatabase
 from setproctitle import setproctitle
 
@@ -14,7 +15,7 @@ from frigate.config import FrigateConfig
 from frigate.models import Event, Recordings, RecordingsToDelete, Timeline
 from frigate.record.cleanup import RecordingCleanup
 from frigate.record.maintainer import RecordingMaintainer
-from frigate.types import RecordMetricsTypes
+from frigate.types import FeatureMetricsTypes
 from frigate.util import listen
 
 logger = logging.getLogger(__name__)
@@ -22,8 +23,8 @@ logger = logging.getLogger(__name__)
 
 def manage_recordings(
     config: FrigateConfig,
-    recordings_info_queue: mp.Queue,
+    recordings_info_queue: ff.Queue,
-    process_info: dict[str, RecordMetricsTypes],
+    process_info: dict[str, FeatureMetricsTypes],
 ) -> None:
     stop_event = mp.Event()
 

frigate/stats.py

@@ -262,8 +262,12 @@ def stats_snapshot(
     for name, detector in stats_tracking["detectors"].items():
         pid = detector.detect_process.pid if detector.detect_process else None
         stats["detectors"][name] = {
-            "inference_speed": round(detector.avg_inference_speed.value * 1000, 2),
+            "inference_speed": round(detector.avg_inference_speed.value * 1000, 2),  # type: ignore[attr-defined]
-            "detection_start": detector.detection_start.value,
+            # issue https://github.com/python/typeshed/issues/8799
+            # from mypy 0.981 onwards
+            "detection_start": detector.detection_start.value,  # type: ignore[attr-defined]
+            # issue https://github.com/python/typeshed/issues/8799
+            # from mypy 0.981 onwards
             "pid": pid,
         }
     stats["detection_fps"] = round(total_detection_fps, 2)

frigate/timeline.py

@@ -3,9 +3,10 @@
 import logging
 import queue
 import threading
-from multiprocessing.queues import Queue
 from multiprocessing.synchronize import Event as MpEvent
 
+from faster_fifo import Queue
+
 from frigate.config import FrigateConfig
 from frigate.events.maintainer import EventTypeEnum
 from frigate.models import Timeline

frigate/types.py

@@ -1,8 +1,9 @@
 from multiprocessing.context import Process
-from multiprocessing.queues import Queue
 from multiprocessing.sharedctypes import Synchronized
 from typing import Optional, TypedDict
 
+from faster_fifo import Queue
+
 from frigate.object_detection import ObjectDetectProcess
 
 
@@ -24,7 +25,8 @@ class CameraMetricsTypes(TypedDict):
     skipped_fps: Synchronized
 
 
-class RecordMetricsTypes(TypedDict):
+class FeatureMetricsTypes(TypedDict):
+    audio_enabled: Synchronized
     record_enabled: Synchronized
 
 

frigate/util.py

@@ -1,18 +1,22 @@
 import copy
+import ctypes
 import datetime
 import json
 import logging
+import multiprocessing
 import os
 import re
 import shlex
 import signal
 import subprocess as sp
+import time
 import traceback
 import urllib.parse
 from abc import ABC, abstractmethod
 from collections import Counter
 from collections.abc import Mapping
 from multiprocessing import shared_memory
+from queue import Empty, Full
 from typing import Any, AnyStr, Optional, Tuple
 
 import cv2
@@ -21,8 +25,12 @@ import psutil
 import py3nvml.py3nvml as nvml
 import pytz
 import yaml
+
 from PIL import Image
 
+from faster_fifo import DEFAULT_CIRCULAR_BUFFER_SIZE, DEFAULT_TIMEOUT
+from faster_fifo import Queue as FFQueue
+
 from frigate.const import REGEX_HTTP_CAMERA_USER_PASS, REGEX_RTSP_CAMERA_USER_PASS
 
 logger = logging.getLogger(__name__)
@@ -629,34 +637,42 @@ def restart_frigate():
 
 
 class EventsPerSecond:
-    def __init__(self, max_events=1000):
+    def __init__(self, max_events=1000, last_n_seconds=10):
         self._start = None
         self._max_events = max_events
+        self._last_n_seconds = last_n_seconds
         self._timestamps = []
 
     def start(self):
         self._start = datetime.datetime.now().timestamp()
 
     def update(self):
+        now = datetime.datetime.now().timestamp()
         if self._start is None:
-            self.start()
+            self._start = now
-        self._timestamps.append(datetime.datetime.now().timestamp())
+        self._timestamps.append(now)
         # truncate the list when it goes 100 over the max_size
         if len(self._timestamps) > self._max_events + 100:
             self._timestamps = self._timestamps[(1 - self._max_events) :]
+        self.expire_timestamps(now)
 
-    def eps(self, last_n_seconds=10):
+    def eps(self):
-        if self._start is None:
-            self.start()
-        # compute the (approximate) events in the last n seconds
         now = datetime.datetime.now().timestamp()
-        seconds = min(now - self._start, last_n_seconds)
+        if self._start is None:
+            self._start = now
+        # compute the (approximate) events in the last n seconds
+        self.expire_timestamps(now)
+        seconds = min(now - self._start, self._last_n_seconds)
         # avoid divide by zero
         if seconds == 0:
             seconds = 1
-        return (
+        return len(self._timestamps) / seconds
-            len([t for t in self._timestamps if t > (now - last_n_seconds)]) / seconds
+
-        )
+    # remove aged out timestamps
+    def expire_timestamps(self, now):
+        threshold = now - self._last_n_seconds
+        while self._timestamps and self._timestamps[0] < threshold:
+            del self._timestamps[0]
 
 
 def print_stack(sig, frame):
@@ -1189,3 +1205,47 @@ def get_video_properties(url, get_duration=False):
         result["height"] = round(height)
 
     return result
+
+
+class LimitedQueue(FFQueue):
+    def __init__(
+        self,
+        maxsize=0,
+        max_size_bytes=DEFAULT_CIRCULAR_BUFFER_SIZE,
+        loads=None,
+        dumps=None,
+    ):
+        super().__init__(max_size_bytes=max_size_bytes, loads=loads, dumps=dumps)
+        self.maxsize = maxsize
+        self.size = multiprocessing.RawValue(
+            ctypes.c_int, 0
+        )  # Add a counter for the number of items in the queue
+
+    def put(self, x, block=True, timeout=DEFAULT_TIMEOUT):
+        if self.maxsize > 0 and self.size.value >= self.maxsize:
+            if block:
+                start_time = time.time()
+                while self.size.value >= self.maxsize:
+                    remaining = timeout - (time.time() - start_time)
+                    if remaining <= 0.0:
+                        raise Full
+                    time.sleep(min(remaining, 0.1))
+            else:
+                raise Full
+        self.size.value += 1
+        return super().put(x, block=block, timeout=timeout)
+
+    def get(self, block=True, timeout=DEFAULT_TIMEOUT):
+        if self.size.value <= 0 and not block:
+            raise Empty
+        self.size.value -= 1
+        return super().get(block=block, timeout=timeout)
+
+    def qsize(self):
+        return self.size
+
+    def empty(self):
+        return self.qsize() == 0
+
+    def full(self):
+        return self.qsize() == self.maxsize

frigate/video.py

@@ -11,11 +11,12 @@ import time
 from collections import defaultdict
 
 import cv2
+import faster_fifo as ff
 import numpy as np
 from setproctitle import setproctitle
 
-from frigate.config import CameraConfig, DetectConfig
+from frigate.config import CameraConfig, DetectConfig, ModelConfig
-from frigate.const import CACHE_DIR
+from frigate.const import ALL_ATTRIBUTE_LABELS, ATTRIBUTE_LABEL_MAP, CACHE_DIR
 from frigate.detectors.detector_config import PixelFormatEnum
 from frigate.log import LogPipe
 from frigate.motion import MotionDetector
@@ -95,7 +96,17 @@ def filtered(obj, objects_to_track, object_filters):
     return False
 
 
-def create_tensor_input(frame, model_config, region):
+def get_min_region_size(model_config: ModelConfig) -> int:
+    """Get the min region size and ensure it is divisible by 4."""
+    half = int(max(model_config.height, model_config.width) / 2)
+
+    if half % 4 == 0:
+        return half
+
+    return int((half + 3) / 4) * 4
+
+
+def create_tensor_input(frame, model_config: ModelConfig, region):
     if model_config.input_pixel_format == PixelFormatEnum.rgb:
         cropped_frame = yuv_region_2_rgb(frame, region)
     elif model_config.input_pixel_format == PixelFormatEnum.bgr:
@@ -172,7 +183,7 @@ def capture_frames(
     skipped_eps.start()
     while True:
         fps.value = frame_rate.eps()
-        skipped_eps.eps()
+        skipped_fps.value = skipped_eps.eps()
 
         current_frame.value = datetime.datetime.now().timestamp()
         frame_name = f"{camera_name}{current_frame.value}"
@@ -215,6 +226,7 @@ class CameraWatchdog(threading.Thread):
         config: CameraConfig,
         frame_queue,
         camera_fps,
+        skipped_fps,
         ffmpeg_pid,
         stop_event,
     ):
@@ -227,11 +239,13 @@ class CameraWatchdog(threading.Thread):
         self.logpipe = LogPipe(f"ffmpeg.{self.camera_name}.detect")
         self.ffmpeg_other_processes: list[dict[str, any]] = []
         self.camera_fps = camera_fps
+        self.skipped_fps = skipped_fps
         self.ffmpeg_pid = ffmpeg_pid
         self.frame_queue = frame_queue
         self.frame_shape = self.config.frame_shape_yuv
         self.frame_size = self.frame_shape[0] * self.frame_shape[1]
         self.stop_event = stop_event
+        self.sleeptime = self.config.ffmpeg.retry_interval
 
     def run(self):
         self.start_ffmpeg_detect()
@@ -251,8 +265,8 @@ class CameraWatchdog(threading.Thread):
                 }
             )
 
-        time.sleep(10)
+        time.sleep(self.sleeptime)
-        while not self.stop_event.wait(10):
+        while not self.stop_event.wait(self.sleeptime):
             now = datetime.datetime.now().timestamp()
 
             if not self.capture_thread.is_alive():
@@ -346,6 +360,7 @@ class CameraWatchdog(threading.Thread):
             self.frame_shape,
             self.frame_queue,
             self.camera_fps,
+            self.skipped_fps,
             self.stop_event,
         )
         self.capture_thread.start()
@@ -376,7 +391,14 @@ class CameraWatchdog(threading.Thread):
 
 class CameraCapture(threading.Thread):
     def __init__(
-        self, camera_name, ffmpeg_process, frame_shape, frame_queue, fps, stop_event
+        self,
+        camera_name,
+        ffmpeg_process,
+        frame_shape,
+        frame_queue,
+        fps,
+        skipped_fps,
+        stop_event,
     ):
         threading.Thread.__init__(self)
         self.name = f"capture:{camera_name}"
@@ -385,14 +407,13 @@ class CameraCapture(threading.Thread):
         self.frame_queue = frame_queue
         self.fps = fps
         self.stop_event = stop_event
-        self.skipped_fps = EventsPerSecond()
+        self.skipped_fps = skipped_fps
         self.frame_manager = SharedMemoryFrameManager()
         self.ffmpeg_process = ffmpeg_process
         self.current_frame = mp.Value("d", 0.0)
         self.last_frame = 0
 
     def run(self):
-        self.skipped_fps.start()
         capture_frames(
             self.ffmpeg_process,
             self.camera_name,
@@ -424,6 +445,7 @@ def capture_camera(name, config: CameraConfig, process_info):
         config,
         frame_queue,
         process_info["camera_fps"],
+        process_info["skipped_fps"],
         process_info["ffmpeg_pid"],
         stop_event,
     )
@@ -706,15 +728,15 @@ def get_consolidated_object_detections(detected_object_groups):
 
 def process_frames(
     camera_name: str,
-    frame_queue: mp.Queue,
+    frame_queue: ff.Queue,
     frame_shape,
-    model_config,
+    model_config: ModelConfig,
     detect_config: DetectConfig,
     frame_manager: FrameManager,
     motion_detector: MotionDetector,
     object_detector: RemoteObjectDetector,
     object_tracker: ObjectTracker,
-    detected_objects_queue: mp.Queue,
+    detected_objects_queue: ff.Queue,
     process_info: dict,
     objects_to_track: list[str],
     object_filters,
@@ -723,14 +745,6 @@ def process_frames(
     stop_event,
     exit_on_empty: bool = False,
 ):
-    # attribute labels are not tracked and are not assigned regions
-    attribute_label_map = {
-        "person": ["face", "amazon"],
-        "car": ["ups", "fedex", "amazon", "license_plate"],
-    }
-    all_attribute_labels = [
-        item for sublist in attribute_label_map.values() for item in sublist
-    ]
     fps = process_info["process_fps"]
     detection_fps = process_info["detection_fps"]
     current_frame_time = process_info["detection_frame"]
@@ -740,7 +754,7 @@ def process_frames(
 
     startup_scan_counter = 0
 
-    region_min_size = int(max(model_config.height, model_config.width) / 2)
+    region_min_size = get_min_region_size(model_config)
 
     while not stop_event.is_set():
         if exit_on_empty and frame_queue.empty():
@@ -906,7 +920,7 @@ def process_frames(
                 tracked_detections = [
                     d
                     for d in consolidated_detections
-                    if d[0] not in all_attribute_labels
+                    if d[0] not in ALL_ATTRIBUTE_LABELS
                 ]
                 # now that we have refined our detections, we need to track objects
                 object_tracker.match_and_update(frame_time, tracked_detections)
@@ -916,7 +930,7 @@ def process_frames(
 
         # group the attribute detections based on what label they apply to
         attribute_detections = {}
-        for label, attribute_labels in attribute_label_map.items():
+        for label, attribute_labels in ATTRIBUTE_LABEL_MAP.items():
             attribute_detections[label] = [
                 d for d in consolidated_detections if d[0] in attribute_labels
             ]

frigate/watchdog.py

@@ -24,7 +24,9 @@ class FrigateWatchdog(threading.Thread):
 
             # check the detection processes
             for detector in self.detectors.values():
-                detection_start = detector.detection_start.value
+                detection_start = detector.detection_start.value  # type: ignore[attr-defined]
+                # issue https://github.com/python/typeshed/issues/8799
+                # from mypy 0.981 onwards
                 if detection_start > 0.0 and now - detection_start > 10:
                     logger.info(
                         "Detection appears to be stuck. Restarting detection process..."

requirements-wheels.txt

@@ -1,14 +1,15 @@
 click == 8.1.*
 Flask == 2.3.*
+faster-fifo == 1.4.*
 imutils == 0.5.*
 matplotlib == 3.7.*
-mypy == 0.942
+mypy == 1.4.1
 numpy == 1.23.*
 onvif_zeep == 0.2.12
-opencv-python-headless == 4.5.5.*
+opencv-python-headless == 4.7.0.*
 paho-mqtt == 1.6.*
 peewee == 3.16.*
-peewee_migrate == 1.10.*
+peewee_migrate == 1.11.*
 psutil == 5.9.*
 pydantic == 1.10.*
 git+https://github.com/fbcotter/py3nvml#egg=py3nvml

web/__test__/handlers.js

@@ -16,6 +16,7 @@ export const handlers = [
           front: {
             name: 'front',
             objects: { track: ['taco', 'cat', 'dog'] },
+            audio: { enabled: false, enabled_in_config: false },
             record: { enabled: true, enabled_in_config: true },
             detect: { width: 1280, height: 720 },
             snapshots: {},
@@ -25,6 +26,7 @@ export const handlers = [
           side: {
             name: 'side',
             objects: { track: ['taco', 'cat', 'dog'] },
+            audio: { enabled: false, enabled_in_config: false },
             record: { enabled: false, enabled_in_config: true },
             detect: { width: 1280, height: 720 },
             snapshots: {},

web/index.html

@@ -8,7 +8,7 @@
     <link rel="apple-touch-icon" sizes="180x180" href="/images/apple-touch-icon.png" />
     <link rel="icon" type="image/png" sizes="32x32" href="/images/favicon-32x32.png" />
     <link rel="icon" type="image/png" sizes="16x16" href="/images/favicon-16x16.png" />
-    <link rel="icon" type="image/svg+xml" href="/images/favicon.svg">
+    <link rel="icon" type="image/svg+xml" href="/images/favicon.svg" />
     <link rel="manifest" href="/site.webmanifest" />
     <link rel="mask-icon" href="/images/favicon.svg" color="#3b82f7" />
     <meta name="msapplication-TileColor" content="#3b82f7" />

web/package.json

@@ -38,8 +38,8 @@
     "@testing-library/user-event": "^14.4.3",
     "@typescript-eslint/eslint-plugin": "^5.59.1",
     "@typescript-eslint/parser": "^5.59.1",
-    "@vitest/coverage-c8": "^0.31.0",
+    "@vitest/coverage-v8": "^0.32.2",
-    "@vitest/ui": "^0.31.0",
+    "@vitest/ui": "^0.32.2",
     "autoprefixer": "^10.4.14",
     "eslint": "^8.39.0",
     "eslint-config-preact": "^1.3.0",
@@ -53,6 +53,6 @@
     "tailwindcss": "^3.3.2",
     "typescript": "^5.0.4",
     "vite": "^4.3.5",
-    "vitest": "^0.31.0"
+    "vitest": "^0.32.2"
   }
 }

web/src/api/__tests__/ws.test.jsx

@@ -113,8 +113,8 @@ describe('WsProvider', () => {
     vi.spyOn(Date, 'now').mockReturnValue(123456);
     const config = {
       cameras: {
-        front: { name: 'front', detect: { enabled: true }, record: { enabled: false }, snapshots: { enabled: true } },
+        front: { name: 'front', detect: { enabled: true }, record: { enabled: false }, snapshots: { enabled: true }, audio: { enabled: false } },
-        side: { name: 'side', detect: { enabled: false }, record: { enabled: false }, snapshots: { enabled: false } },
+        side: { name: 'side', detect: { enabled: false }, record: { enabled: false }, snapshots: { enabled: false }, audio: { enabled: false } },
       },
     };
     render(

web/src/api/ws.jsx

@@ -41,10 +41,11 @@ export function WsProvider({
 
   useEffect(() => {
     Object.keys(config.cameras).forEach((camera) => {
-      const { name, record, detect, snapshots } = config.cameras[camera];
+      const { name, record, detect, snapshots, audio } = config.cameras[camera];
       dispatch({ topic: `${name}/recordings/state`, payload: record.enabled ? 'ON' : 'OFF', retain: false });
       dispatch({ topic: `${name}/detect/state`, payload: detect.enabled ? 'ON' : 'OFF', retain: false });
       dispatch({ topic: `${name}/snapshots/state`, payload: snapshots.enabled ? 'ON' : 'OFF', retain: false });
+      dispatch({ topic: `${name}/audio/state`, payload: audio.enabled ? 'ON' : 'OFF', retain: false });
     });
   }, [config]);
 
@@ -120,6 +121,15 @@ export function useSnapshotsState(camera) {
   return { payload, send, connected };
 }
 
+export function useAudioState(camera) {
+  const {
+    value: { payload },
+    send,
+    connected,
+  } = useWs(`${camera}/audio/state`, `${camera}/audio/set`);
+  return { payload, send, connected };
+}
+
 export function usePtzCommand(camera) {
   const {
     value: { payload },

web/src/components/CameraImage.jsx

@@ -28,13 +28,18 @@ export default function CameraImage({ camera, onload, searchParams = '', stretch
 
   const scaledHeight = useMemo(() => {
     const scaledHeight = Math.floor(availableWidth / aspectRatio);
-    return stretch ? scaledHeight : Math.min(scaledHeight, height);
+    const finalHeight = stretch ? scaledHeight : Math.min(scaledHeight, height);
+
+    if (finalHeight > 0) {
+      return finalHeight;
+    }
+
+    return 100;
   }, [availableWidth, aspectRatio, height, stretch]);
-  const scaledWidth = useMemo(() => Math.ceil(scaledHeight * aspectRatio - scrollBarWidth), [
+  const scaledWidth = useMemo(
-    scaledHeight,
+    () => Math.ceil(scaledHeight * aspectRatio - scrollBarWidth),
-    aspectRatio,
+    [scaledHeight, aspectRatio, scrollBarWidth]
-    scrollBarWidth,
+  );
-  ]);
 
   const img = useMemo(() => new Image(), []);
   img.onload = useCallback(
@@ -58,18 +63,16 @@ export default function CameraImage({ camera, onload, searchParams = '', stretch
 
   return (
     <div className="relative w-full" ref={containerRef}>
-      {
+      {enabled ? (
-        (enabled) ?
         <canvas data-testid="cameraimage-canvas" height={scaledHeight} ref={canvasRef} width={scaledWidth} />
-          : <div class="text-center pt-6">Camera is disabled in config, no stream or snapshot available!</div>
+      ) : (
-      }
+        <div class="text-center pt-6">Camera is disabled in config, no stream or snapshot available!</div>
-      {
+      )}
-        (!hasLoaded && enabled) ? (
+      {!hasLoaded && enabled ? (
         <div className="absolute inset-0 flex justify-center" style={`height: ${scaledHeight}px`}>
           <ActivityIndicator />
         </div>
-        ) : null
+      ) : null}
-      }
+    </div>
-    </div >
   );
 }

web/src/components/TimeAgo.tsx

@@ -25,9 +25,9 @@ const timeAgo = ({ time, currentTime = new Date(), dense = false }: IProp): stri
   const elapsedTime: number = currentTime.getTime() - pastTime.getTime();
 
   const timeUnits: TimeUnit[] = [
-    { unit: 'ye', full: 'year', value: 31536000 },
+    { unit: 'yr', full: 'year', value: 31536000 },
     { unit: 'mo', full: 'month', value: 0 },
-    { unit: 'day', full: 'day', value: 86400 },
+    { unit: 'd', full: 'day', value: 86400 },
     { unit: 'h', full: 'hour', value: 3600 },
     { unit: 'm', full: 'minute', value: 60 },
     { unit: 's', full: 'second', value: 1 },
@@ -58,11 +58,11 @@ const timeAgo = ({ time, currentTime = new Date(), dense = false }: IProp): stri
 
       if (monthDiff > 0) {
         const unitAmount = monthDiff;
-        return `${unitAmount}${dense ? timeUnits[i].unit[0] : ` ${timeUnits[i].full}`}${dense ? '' : 's'} ago`;
+        return `${unitAmount}${dense ? timeUnits[i].unit : ` ${timeUnits[i].full}`}${dense ? '' : 's'} ago`;
       }
     } else if (elapsed >= timeUnits[i].value) {
       const unitAmount: number = Math.floor(elapsed / timeUnits[i].value);
-      return `${unitAmount}${dense ? timeUnits[i].unit[0] : ` ${timeUnits[i].full}`}${dense ? '' : 's'} ago`;
+      return `${unitAmount}${dense ? timeUnits[i].unit : ` ${timeUnits[i].full}`}${dense ? '' : 's'} ago`;
     }
   }
   return 'Invalid Time';

web/src/components/TimelineEventOverlay.jsx

@@ -0,0 +1,65 @@
+import { Fragment, h } from 'preact';
+import { useState } from 'preact/hooks';
+
+export default function TimelineEventOverlay({ eventOverlay, cameraConfig }) {
+  const boxLeftEdge = Math.round(eventOverlay.data.box[0] * 100);
+  const boxTopEdge = Math.round(eventOverlay.data.box[1] * 100);
+  const boxRightEdge = Math.round((1 - eventOverlay.data.box[2] - eventOverlay.data.box[0]) * 100);
+  const boxBottomEdge = Math.round((1 - eventOverlay.data.box[3] - eventOverlay.data.box[1]) * 100);
+
+  const [isHovering, setIsHovering] = useState(false);
+  const getHoverStyle = () => {
+    if (boxLeftEdge < 15) {
+      // show object stats on right side
+      return {
+        left: `${boxLeftEdge + eventOverlay.data.box[2] * 100 + 1}%`,
+        top: `${boxTopEdge}%`,
+      };
+    }
+
+    return {
+      right: `${boxRightEdge + eventOverlay.data.box[2] * 100 + 1}%`,
+      top: `${boxTopEdge}%`,
+    };
+  };
+
+  const getObjectArea = () => {
+    const width = eventOverlay.data.box[2] * cameraConfig.detect.width;
+    const height = eventOverlay.data.box[3] * cameraConfig.detect.height;
+    return Math.round(width * height);
+  };
+
+  const getObjectRatio = () => {
+    const width = eventOverlay.data.box[2] * cameraConfig.detect.width;
+    const height = eventOverlay.data.box[3] * cameraConfig.detect.height;
+    return Math.round(100 * (width / height)) / 100;
+  };
+
+  return (
+    <Fragment>
+      <div
+        className="absolute border-4 border-red-600"
+        onMouseEnter={() => setIsHovering(true)}
+        onMouseLeave={() => setIsHovering(false)}
+        onTouchStart={() => setIsHovering(true)}
+        onTouchEnd={() => setIsHovering(false)}
+        style={{
+          left: `${boxLeftEdge}%`,
+          top: `${boxTopEdge}%`,
+          right: `${boxRightEdge}%`,
+          bottom: `${boxBottomEdge}%`,
+        }}
+      >
+        {eventOverlay.class_type == 'entered_zone' ? (
+          <div className="absolute w-2 h-2 bg-yellow-500 left-[50%] -translate-x-1/2 translate-y-3/4 bottom-0" />
+        ) : null}
+      </div>
+      {isHovering && (
+        <div className="absolute bg-white dark:bg-slate-800 p-4 block dark:text-white text-lg" style={getHoverStyle()}>
+          <div>{`Area: ${getObjectArea()} px`}</div>
+          <div>{`Ratio: ${getObjectRatio()}`}</div>
+        </div>
+      )}
+    </Fragment>
+  );
+}

web/src/components/Tooltip.jsx

@@ -29,7 +29,7 @@ export default function Tooltip({ relativeTo, text }) {
       let newLeft = left - Math.round(tipWidth / 2);
       // too far right
       if (newLeft + tipWidth + TIP_SPACE > windowWidth - window.scrollX) {
-        newLeft = left - tipWidth - TIP_SPACE;
+        newLeft = Math.max(0, left - tipWidth - TIP_SPACE);
         newTop = top - Math.round(tipHeight / 2);
       }
       // too far left

web/src/icons/Audio.jsx

@@ -0,0 +1,36 @@
+import { h } from 'preact';
+import { memo } from 'preact/compat';
+
+export function Snapshot({ className = 'h-6 w-6', stroke = 'currentColor', onClick = () => {} }) {
+  return (
+    <svg
+      xmlns="http://www.w3.org/2000/svg"
+      className={className}
+      fill="none"
+      viewBox="0 0 32 32"
+      stroke={stroke}
+      onClick={onClick}
+    >
+      <path
+        stroke-linecap="round"
+        stroke-linejoin="round"
+        stroke-width="2"
+        d="M18 30v-2a10.011 10.011 0 0010-10h2a12.013 12.013 0 01-12 12z"
+      />
+      <path
+        stroke-linecap="round"
+        stroke-linejoin="round"
+        stroke-width="2"
+        d="M18 22v-2a2.002 2.002 0 002-2h2a4.004 4.004 0 01-4 4zM10 2a9.01 9.01 0 00-9 9h2a7 7 0 0114 0 7.09 7.09 0 01-3.501 6.135l-.499.288v3.073a2.935 2.935 0 01-.9 2.151 4.182 4.182 0 01-4.633 1.03A4.092 4.092 0 015 20H3a6.116 6.116 0 003.67 5.512 5.782 5.782 0 002.314.486 6.585 6.585 0 004.478-1.888A4.94 4.94 0 0015 20.496v-1.942A9.108 9.108 0 0019 11a9.01 9.01 0 00-9-9z"
+      />
+      <path
+        stroke-linecap="round"
+        stroke-linejoin="round"
+        stroke-width="2"
+        d="M9.28 8.082A3.006 3.006 0 0113 11h2a4.979 4.979 0 00-1.884-3.911 5.041 5.041 0 00-4.281-.957 4.95 4.95 0 00-3.703 3.703 5.032 5.032 0 002.304 5.458A3.078 3.078 0 019 17.924V20h2v-2.077a5.06 5.06 0 00-2.537-4.346 3.002 3.002 0 01.817-5.494z"
+      />
+    </svg>
+  );
+}
+
+export default memo(Snapshot);

web/src/routes/Camera.jsx

@@ -22,6 +22,7 @@ const emptyObject = Object.freeze({});
 
 export default function Camera({ camera }) {
   const { data: config } = useSWR('config');
+  const { data: trackedLabels } = useSWR(['labels', { camera }]);
   const apiHost = useApiHost();
   const [showSettings, setShowSettings] = useState(false);
   const [viewMode, setViewMode] = useState('live');
@@ -121,7 +122,9 @@ export default function Camera({ camera }) {
             <div className="max-w-5xl">
               <video-stream
                 mode="mse"
-                src={new URL(`${baseUrl.replace(/^http/, 'ws')}live/webrtc/api/ws?src=${cameraConfig.live.stream_name}`)}
+                src={
+                  new URL(`${baseUrl.replace(/^http/, 'ws')}live/webrtc/api/ws?src=${cameraConfig.live.stream_name}`)
+                }
               />
             </div>
           </Fragment>
@@ -203,7 +206,7 @@ export default function Camera({ camera }) {
       <div className="space-y-4">
         <Heading size="sm">Tracked objects</Heading>
         <div className="flex flex-wrap justify-start">
-          {cameraConfig.objects.track.map((objectType) => (
+          {(trackedLabels || []).map((objectType) => (
             <Card
               className="mb-4 mr-4"
               key={objectType}

web/src/routes/Cameras.jsx

@@ -2,10 +2,11 @@ import { h, Fragment } from 'preact';
 import ActivityIndicator from '../components/ActivityIndicator';
 import Card from '../components/Card';
 import CameraImage from '../components/CameraImage';
+import AudioIcon from '../icons/Audio';
 import ClipIcon from '../icons/Clip';
 import MotionIcon from '../icons/Motion';
 import SnapshotIcon from '../icons/Snapshot';
-import { useDetectState, useRecordingsState, useSnapshotsState } from '../api/ws';
+import { useAudioState, useDetectState, useRecordingsState, useSnapshotsState } from '../api/ws';
 import { useMemo } from 'preact/hooks';
 import useSWR from 'swr';
 
@@ -43,6 +44,7 @@ function Camera({ name, config }) {
   const { payload: detectValue, send: sendDetect } = useDetectState(name);
   const { payload: recordValue, send: sendRecordings } = useRecordingsState(name);
   const { payload: snapshotValue, send: sendSnapshots } = useSnapshotsState(name);
+  const { payload: audioValue, send: sendAudio } = useAudioState(name);
   const href = `/cameras/${name}`;
   const buttons = useMemo(() => {
     return [
@@ -50,10 +52,9 @@ function Camera({ name, config }) {
       { name: 'Recordings', href: `/recording/${name}` },
     ];
   }, [name]);
-  const cleanName = useMemo(
+  const cleanName = useMemo(() => {
-    () => { return `${name.replaceAll('_', ' ')}` },
+    return `${name.replaceAll('_', ' ')}`;
-    [name]
+  }, [name]);
-  );
   const icons = useMemo(
     () => [
       {
@@ -65,7 +66,9 @@ function Camera({ name, config }) {
         },
       },
       {
-        name: config.record.enabled_in_config ? `Toggle recordings ${recordValue === 'ON' ? 'off' : 'on'}` : 'Recordings must be enabled in the config to be turned on in the UI.',
+        name: config.record.enabled_in_config
+          ? `Toggle recordings ${recordValue === 'ON' ? 'off' : 'on'}`
+          : 'Recordings must be enabled in the config to be turned on in the UI.',
         icon: ClipIcon,
         color: config.record.enabled_in_config ? (recordValue === 'ON' ? 'blue' : 'gray') : 'red',
         onClick: () => {
@@ -82,11 +85,27 @@ function Camera({ name, config }) {
           sendSnapshots(snapshotValue === 'ON' ? 'OFF' : 'ON', true);
         },
       },
-    ],
+      config.audio.enabled_in_config
-    [config, detectValue, sendDetect, recordValue, sendRecordings, snapshotValue, sendSnapshots]
+        ? {
+          name: `Toggle audio detection ${audioValue === 'ON' ? 'off' : 'on'}`,
+          icon: AudioIcon,
+          color: audioValue === 'ON' ? 'blue' : 'gray',
+          onClick: () => {
+            sendAudio(audioValue === 'ON' ? 'OFF' : 'ON', true);
+          },
+        }
+        : null,
+    ].filter((button) => button != null),
+    [config, audioValue, sendAudio, detectValue, sendDetect, recordValue, sendRecordings, snapshotValue, sendSnapshots]
   );
 
   return (
-    <Card buttons={buttons} href={href} header={cleanName} icons={icons} media={<CameraImage camera={name} stretch />} />
+    <Card
+      buttons={buttons}
+      href={href}
+      header={cleanName}
+      icons={icons}
+      media={<CameraImage camera={name} stretch />}
+    />
   );
 }

web/src/routes/Events.jsx

@@ -29,6 +29,7 @@ import { formatUnixTimestampToDateTime, getDurationFromTimestamps } from '../uti
 import TimeAgo from '../components/TimeAgo';
 import Timepicker from '../components/TimePicker';
 import TimelineSummary from '../components/TimelineSummary';
+import TimelineEventOverlay from '../components/TimelineEventOverlay';
 
 const API_LIMIT = 25;
 
@@ -106,6 +107,7 @@ export default function Events({ path, ...props }) {
 
   const { data: config } = useSWR('config');
 
+  const { data: allLabels } = useSWR(['labels']);
   const { data: allSubLabels } = useSWR(['sub_labels', { split_joined: 1 }]);
 
   const filterValues = useMemo(
@@ -120,15 +122,10 @@ export default function Events({ path, ...props }) {
           .filter((value, i, self) => self.indexOf(value) === i),
         'None',
       ],
-      labels: Object.values(config?.cameras || {})
+      labels: Object.values(allLabels || {}),
-        .reduce((memo, camera) => {
-          memo = memo.concat(camera?.objects?.track || []);
-          return memo;
-        }, config?.objects?.track || [])
-        .filter((value, i, self) => self.indexOf(value) === i),
       sub_labels: (allSubLabels || []).length > 0 ? [...Object.values(allSubLabels), 'None'] : [],
     }),
-    [config, allSubLabels]
+    [config, allLabels, allSubLabels]
   );
 
   const onSave = async (e, eventId, save) => {
@@ -721,23 +718,10 @@ export default function Events({ path, ...props }) {
                                   }}
                                 >
                                   {eventOverlay ? (
-                                    <div
+                                    <TimelineEventOverlay
-                                      className="absolute border-4 border-red-600"
+                                      eventOverlay={eventOverlay}
-                                      style={{
+                                      cameraConfig={config.cameras[event.camera]}
-                                        left: `${Math.round(eventOverlay.data.box[0] * 100)}%`,
+                                    />
-                                        top: `${Math.round(eventOverlay.data.box[1] * 100)}%`,
-                                        right: `${Math.round(
-                                          (1 - eventOverlay.data.box[2] - eventOverlay.data.box[0]) * 100
-                                        )}%`,
-                                        bottom: `${Math.round(
-                                          (1 - eventOverlay.data.box[3] - eventOverlay.data.box[1]) * 100
-                                        )}%`,
-                                      }}
-                                    >
-                                      {eventOverlay.class_type == 'entered_zone' ? (
-                                        <div className="absolute w-2 h-2 bg-yellow-500 left-[50%] -translate-x-1/2 translate-y-3/4 bottom-0" />
-                                      ) : null}
-                                    </div>
                                   ) : null}
                                 </VideoPlayer>
                               </div>

web/src/routes/Export.jsx

@@ -18,9 +18,9 @@ export default function Export() {
   const localISODate = localDate.toISOString().split('T')[0];
 
   const [startDate, setStartDate] = useState(localISODate);
-  const [startTime, setStartTime] = useState("00:00");
+  const [startTime, setStartTime] = useState('00:00');
   const [endDate, setEndDate] = useState(localISODate);
-  const [endTime, setEndTime] = useState("23:59");
+  const [endTime, setEndTime] = useState('23:59');
 
   const onHandleExport = () => {
     if (camera == 'select') {
@@ -33,8 +33,6 @@ export default function Export() {
       return;
     }
 
-    
-
     if (!startDate || !startTime || !endDate || !endTime) {
       setMessage({ text: 'A start and end time needs to be selected', error: true });
       return;
@@ -48,12 +46,13 @@ export default function Export() {
       return;
     }
 
-    axios.post(`export/${camera}/start/${start}/end/${end}`, { playback })
+    axios
+      .post(`export/${camera}/start/${start}/end/${end}`, { playback })
       .then(() => {
         setMessage({ text: 'Successfully started export. View the file in the /exports folder.', error: false });
       })
       .catch((error) => {
-        setMessage({ text: 'Failed to start export: '+error.response.data.message, error: true });
+        setMessage({ text: `Failed to start export: ${error.response.data.message}`, error: true });
       });
   };
 
@@ -93,13 +92,37 @@ export default function Export() {
         <Heading className="py-2" size="sm">
           From:
         </Heading>
-        <input className="dark:bg-slate-800" id="startDate" type="date" value={startDate} onChange={(e) => setStartDate(e.target.value)}/>
+        <input
-        <input className="dark:bg-slate-800" id="startTime" type="time" value={startTime} onChange={(e) => setStartTime(e.target.value)}/>
+          className="dark:bg-slate-800"
+          id="startDate"
+          type="date"
+          value={startDate}
+          onChange={(e) => setStartDate(e.target.value)}
+        />
+        <input
+          className="dark:bg-slate-800"
+          id="startTime"
+          type="time"
+          value={startTime}
+          onChange={(e) => setStartTime(e.target.value)}
+        />
         <Heading className="py-2" size="sm">
           To:
         </Heading>
-        <input className="dark:bg-slate-800" id="endDate" type="date" value={endDate} onChange={(e) => setEndDate(e.target.value)}/>
+        <input
-        <input className="dark:bg-slate-800" id="endTime" type="time" value={endTime} onChange={(e) => setEndTime(e.target.value)}/>
+          className="dark:bg-slate-800"
+          id="endDate"
+          type="date"
+          value={endDate}
+          onChange={(e) => setEndDate(e.target.value)}
+        />
+        <input
+          className="dark:bg-slate-800"
+          id="endTime"
+          type="time"
+          value={endTime}
+          onChange={(e) => setEndTime(e.target.value)}
+        />
       </div>
       <Button onClick={() => onHandleExport()}>Submit</Button>
     </div>

web/src/routes/System.jsx

@@ -334,7 +334,7 @@ export default function System() {
             <ActivityIndicator />
           ) : (
             <div data-testid="cameras" className="grid grid-cols-1 3xl:grid-cols-3 md:grid-cols-2 gap-4">
-              {cameraNames.map((camera) => (
+              {cameraNames.map((camera) => ( config.cameras[camera]["enabled"] && (
                 <div key={camera} className="dark:bg-gray-800 shadow-md hover:shadow-lg rounded-lg transition-shadow">
                   <div className="capitalize text-lg flex justify-between p-4">
                     <Link href={`/cameras/${camera}`}>{camera.replaceAll('_', ' ')}</Link>
@@ -406,7 +406,7 @@ export default function System() {
                       </Tbody>
                     </Table>
                   </div>
-                </div>
+                </div> )
               ))}
             </div>
           )}