A record-enabled camera whose record stream produces no cache segments
never appears in grouped_recordings, so the per-camera prune in
RecordingMaintainer.move_files() never runs for it. Its
object_recordings_info and audio_recordings_info buffers then grow
without bound until the recording process is OOM-killed (discussion
#23451).
Run a prune every move_files() cycle for cameras absent from
grouped_recordings, dropping entries older than the longest a segment
could still wait in cache before being matched
(MAX_SEGMENTS_IN_CACHE * MAX_SEGMENT_DURATION * 2). Cameras present in
grouped_recordings are left untouched and keep their existing prune.
Add a regression test asserting that an absent camera's stale entries
are dropped (recent ones kept) while a present camera's entries are
left intact.
* display zone names consistently using friendly_name or raw id without transformation
* enforce camera-level access on go2rtc live stream websocket endpoints
* slightly darken bg-card
* change menu label
* move snapshot retain out of advanced fields
* add new ui options for collapsibles
* backend title and description
* remove unused snapshot retention field
* update reference config
* remove further references to snapshots retain.mode
* Implement tool call history keeping
* Refactor to match single message implementation
* Simplify data representation
* Cleanup chat page rendering
* Include system message to not break cache
* Formatting
* Update tests and update .gitignore
* update e2e mock data to remove deprecated fields
* remove scream audio label
scream was never mapped to anything in frigate's custom labelmap, yell is used everywhere
* document common audio labels
* deprecate ffmpeg 5
* language tweak
* add field message to recommend presets instead of manual hwaccel args
* add guidance to docs on choosing a detect fps
* serialize OpenVINO inference per process to prevent concurrent-inference segfault
* clean up
* add max scaling meta to login page
* add more detect section field messages
* fix icon layout in settings field messages
* tweak edit icon color
* republish MQTT switch states when a profile is activated or deactivated
* fix object mask default name when created from Explore tracking details
* tweak annotation offset max in UI
* optimize recordings/unavailable gap detection and drop empty motion activity buckets
* add tests
* refactor motion search
* cleanup dead code and tests
* tweaks
* fix multi-day seeking
* start playback a few seconds before the change so the motion is in view
* add ptz presets and default role widgets
* language tweaks
* fix width in triggers view
* tweak iOS PWA message in notifications settings
* deprecate ui.date_style and ui.time_style
these have been unused since date/time formatting has been pushed to i18n
* add config migrator to remove date_style and time_style
* remove date_style and time_style from reference config
* fix camera list scrolling in state classification wizard on mobile
* improve error parsing and increase skip default
* improve motion search layout to match tracking details
* implement draw and move mode on mobile
* update motion search docs
* language tweaks
* improve tips
* note actions menu
* improve visibility of blurred icon buttons
* add motion search to history actions menu and mobile drawer
* i18n
* use pure css for motion search dialog video
* defer profile restoration until subscribers are connected
* change order of features in mobile review settings drawer
* stabilize chart options to stop ApexCharts updateOptions running on every stats tick
* constrain height of export dialog
* stop audio maintainer when deleting a camera
* run face register and recognize API handlers in threadpool
* resolve global record.export.hwaccel_args to fix phantom camera override
* auto-stop debug replay sessions after 12 hours
* docs tweaks
* add more tips to object classification docs
* tweak language
* Store hwaccel errors with timeout so it can retry
* Add error logs for Intel GPU stats
* add area
---------
Co-authored-by: Nicolas Mowen <nickmowen213@gmail.com>
* add prop to disable id field
* disable id field when editing profile mask/zone
also, disable if the zone name already exists in required_zones or the base config is being edited and the id already exists on a profile
* add backend validation to reject profile-omly masks/zones
* add tests
* update docs
* tweak
* filter motion review by allowed cameras
* filter alertCameras by allowed cameras so the recent alerts query for restricted roles doesn't reference cameras they can't access
* skip data streams in chapter exports to avoid ffmpeg segfault
* formatting
* restrict debug replay UI entry points to admin users
* Adjust default iGPU name when it can't be found
* Fix when model tries to request an invalid camera
* Improve prompt
* add collapsible main nav items in settings
---------
Co-authored-by: Nicolas Mowen <nickmowen213@gmail.com>
* add review padding to explore debug replay api calls
* add semantic search model size widget
disables model_size select with n/a text when an embeddings genai provider is selected
* regenerate zone contours and per-zone filter masks on detect resolution change
* treat null as a clear sentinel in buildOverrides so nullable field edits don't snap back
* extract replay config sheet to new component
* add validation and messages for detect settings
* unlink shm frames when camera is removed
* drop stale shm cache refs when cached segment is too small for requested shape
* skip new-object frame cache write when current_frame is unavailable
* add tests
* use setdefault when adding a new camera
Multiple subscribers in the same process each unpickle the ZMQ payload independently and would otherwise write divergent Python objects to the shared cameras dict — leaving long-lived references (e.g. CameraState.camera_config) pointing at a copy that subsequent in-place mutations like apply_section_update can never reach. setdefault collapses everyone onto the first writer's object so attribute mutations propagate to every consumer in this process.
* rebuild ffmpeg commands on detect update
Rebuild the cached ffmpeg cmd so the next process spawn picks up new resolution/fps. Running cameras keep their existing cmd (ffmpeg_cmds is only read at process startup); replay cameras are recycled by CameraMaintainer to pick up the rebuilt cmd
* drop stale shm cache refs when cached segment size doesn't match requested shape
The cached SharedMemoryFrameManager reference can point at a segment whose
size no longer matches the requested shape — the segment was unlinked and
recreated at a different size in a camera add/remove cycle. This catches
both a resolution increase (cached too small) and a decrease (cached too
large, pointing at an orphaned inode whose stale bytes would otherwise be
misinterpreted at the new shape, producing distorted/miscolored YUV frames).
After reopening, if the OS-level segment still doesn't match the requested
shape we're in a transient mid-recreate state — either the maintainer
hasn't allocated the new segment yet (size too small) or we opened a
pre-recycle segment (size too big). Either way, skip the frame and don't
cache the mismatched ref.
* recycle replay camera on detect update
* discard tracked-object state when detect resolution changes mid-session
When detect resolution changes mid-session every tracked object we hold
was localized against the old pixel grid. Their boxes no longer
correspond to anything in the new frame, and the `end` callback that
fires when their IDs disappear from the new detect process's detections
publishes those stale boxes to consumers (LPR, snapshot crop) that slice
the new frame and crash on empty arrays. Drop the tracked-object state
on a shape change so no stale boxes ever cross the CameraState boundary.
Belt-and-suspenders: also drop any incoming batch whose boxes exceed the
current detect resolution. These are in-flight queue entries from the
pre-recycle detect process that beat the new detect process to the
queue; processing them would re-introduce stale-resolution tracked
objects we just dropped above. The per-camera detect process clamps
legitimate boxes to detect.width-1 / detect.height-1, so any coord
beyond that is unambiguously stale.
* rebuild motion and object filter masks on detect resolution change
Apply the detect update first so frame_shape reflects the new resolution
before we rebuild dependents.
Motion's rasterized_mask is sized to frame_shape at construction. When
detect resolution changes we must rebuild RuntimeMotionConfig so the
mask matches the new frame size; otherwise consumers like the LPR
processor and motion detector hit a shape mismatch when they index
frames with the stale mask.
Same story for per-object filter masks — rebuild RuntimeFilterConfig at
the new frame_shape so the merged global+per-object masks they hold
match what they'll be indexed against.
* republish motion and objects on in-memory detect resize
A detect resolution change also invalidates the rasterized masks on
motion and per-object filters. apply_section_update has rebuilt them at
the new frame_shape; publish them too so other processes replace their
old values.
* add test
* frontend
* add refresh topic for camera maintainer recycle action
The maintainer's recycle branch is doing an action (recycle the camera)
in response to a section-level signal. Introduce a
CameraConfigUpdateEnum.refresh case as an explicit action signal — the
maintainer subscribes to refresh instead of detect, parallel with add
and remove. Publishers fire refresh alongside detect when a recycle is
needed; section-level subscribers keep their existing topic.
Since no main-process subscriber listens for detect anymore, the
refresh handler calls recreate_ffmpeg_cmds() explicitly so the shared
CameraConfig's ffmpeg_cmds is rebuilt before the new subprocesses
spawn.
* factor stale-resolution state drop into a CameraState method
* use monotonic clock for detector inference duration to prevent negative values from wall clock steps
* add ability to set camera's webui_url from camera management pane
* Gemini send thought signature
* Update docs
* copy face and lpr configs from source camera to replay camera
* add guard
* improve dummy camera docs
* remove version number
* fix stale field message after reverting a conditional form field
Routes field-level conditional messages through a dedicated React Context instead of merging them into uiSchema. RJSF's Form keeps state.uiSchema sticky across renders during processPendingChange (formData is updated, uiSchema is not), so a previously injected ui:messages array stays attached to a field even after the triggering condition flips back to false. Context propagation re-runs FieldTemplate directly on every provider value change, sidestepping that staleness.
* add semantic search field message to note that model_size is irrelevant when embeddings provider is selected
---------
Co-authored-by: Nicolas Mowen <nickmowen213@gmail.com>
* filter replay camera from camera selectors
* add face rec and lpr to replay configuration sheet
* add missing config topic subscriptions in embeddings maintainer
* pop replay camera from config object when stopping
* ensure motion masks from source camera are copied to replay
* stop polling debug_replay/status after live_ready
* use vod for constructing replay clips
* render orphaned filter entries as collapsibles instead of the Key/Value editor
* Symlink for various AI files
* change replay confg dialog to platform aware sheet
* change agents title
* fix test
* tweak collapsible
* remove camera ui section in settings
no point to having it anymore with profiles and camera management settings
* fix admin response cache leak to non-admin users via nginx proxy_cache
* add model fetcher endpoint for genai config ui
---------
Co-authored-by: Nicolas Mowen <nickmowen213@gmail.com>
* filter outbound ws broadcasts by per-recipient camera access
* fan out config updates to comms
* tests
* mypy
* allow viewers to use jobstate
* update agent instructions
* remove vitest
* Ensure runtime options are passed
* Add attribute info to prompt when configured
* Move GenAI plugins to dedicated directory
* Migrate prompts to dedicated folder
* Move chat prompts to prompts
* Implement reasoning traces in the UI
* Cleanup
* Make azure a subclass of openai
* Implement reasoning for other providers
* mypy
* Cleanup
* preserve user-set min_score on attribute filters instead of bumping any 0.5 value
use model_fields_set to distinguish "user explicitly set min_score" from "Pydantic applied the generic FilterConfig default of 0.5"
* add config test for attributes
* fix attributes frontend type
* add expanded hidden field context
* extend schema modification
* special case for attributes
* i18n for attributes
* handle dedicated lpr mode
* strip unrendered FilterConfig fields from attribute filter form data to fix validation errors
* start audio transcription post processor when enabled on any camera
* Fetch embed key whenever an error occurs in case the llama server was restarted
* mypy
* add tooltips for colored dots in settings menu
* add ability to reorder cameras from management pane
* add ability to reorder birdseye
* add reordering save text to camera management view
* Include NPU in latency performance hint
* Implement turbo for NPU on object detection
* hide order fields
* drop auto-derived field paths from camera value when unset globally
* use correct field type for export hwaccel args
* add debug replay to detail actions menu
* clarify debug replay in docs
* guard get_current_frame_time against missing camera state
* Implement debug reply from export
* Refactor debug replay to use sources for dynamic playback
* Mypy
* fix debug export replay source timestamp handling
* skip replay cameras in stats immediately
* broadcast debug replay state over ws and buffer pre-OPEN sends
- push debug replay session state over the job_state ws topic so the status bar reacts instantly to start/stop without polling
- fix child-effect-before-parent-effect race in WsProvider that silently dropped initial snapshot requests on cold load
* fix debug replay test hang
---------
Co-authored-by: Nicolas Mowen <nickmowen213@gmail.com>
* add embedded mode to BaseSection so parents can host the save action
* add optional action slot to current Frigate+ model summary
* add w-full to action slot flex wrapper for explicit width contract
* i18n
* merged detectors and model settings view
* fix document title
* Embed detector form in merged settings view
* add detection model card with tabs and custom model embed
* add Frigate+ model selector with filter popover to merged page
* Add mismatch banner and gate save on detector and model compatibility
* Wire atomic save, restart toast, and undo on detectors and model page
* Clear child pending data on undo
* route merged detectors and model view in settings
* trim Frigate+ page to account-only and remove old detection model view
* basic e2e
* Fix unsaved-changes guard, custom path leak, and post-failure cache resync
* Rename to Detectors and model, float Modified badge, use ConfigMessageBanner for mismatch
* Hide Plus/Custom tabs when Frigate+ is not enabled
* Detect active Plus model via model.plus.id instead of path prefix
* Sync state back to snapshot when child form un-modifies and remount on undo
* Always require restart on save since model changes also need one
* Wrap Frigate+ model selector in SplitCardRow with label and description
* rename tab
* update docs
* sync top-level model with default detector's resolved model
when the user doesn't define a top-level `model:` block, `FrigateConfig.model` stayed at pydantic field defaults (320×320, /labelmap.txt) while the per-detector model picked up `DEFAULT_MODEL` for openvino on cpu (300×300, coco_91cl_bkgr.txt introduced in #23127), causing `RemoteObjectDetector` to fail with "buffer is too small for requested array" because the SHM was sized from the per-detector model but mapped using the top-level one. After the detector loop, copy the first detector's resolved model up to `self.model` so both sides agree on dimensions and labelmap
* revert to cpu detector by default
use openvino cpu for new configs only
* add defaults
* sync filter entries with track and listen labels
- Auto-populate `audio.filters` from `audio.listen` instead of the full audio labelmap, matching how `objects.filters` is keyed by `track` (no longer need to populate the full audio labelmap, which was added in #22630)
- Synthesize the matching filter entries in the settings form on load so each track/listen label shows its collapsible after a profile is selected, since the backend's auto-populate only runs at config init
* translate main label for lifecycle description with attribute
* reject restricted go2rtc stream sources when added via api
* add env var check function
* Support token streaming stats
* Propogate streaming token stats to chat calls
* Show token stats for each image
* Add settings to handle token stats and other options
* i18n
* Use select
* Improve mobile layout and spacing
On multi-GPU systems, OpenVINO enumerates devices as "GPU.0", "GPU.1",
etc. rather than a single "GPU". The exact string match in
is_openvino_gpu_npu_available() fails to recognize these suffixed device
names, causing enrichments (face recognition, semantic search) to
silently fall back to CPU-only inference via ONNXModelRunner instead of
using OpenVINOModelRunner on GPU.
Switch from exact match to prefix match so both single-GPU ("GPU") and
multi-GPU ("GPU.0", "GPU.1") device names are correctly detected, along
with any future suffixed variants for NPU and other accelerators.
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* improve scroll handling for non-modal DropdownMenu in classification and face selection dialogs
* clean up
* fix incorrect key capitalization
* fix profile array overrides not replacing base arrays
don't use lodash merge(), it does positional merging and an empty source array doesn't override the destination, and shorter arrays leak destination elements through.
backend is unaffected, so the saved config and actual backend functionality was right
* only show audio debug tab when audio is enabled in config
* move apple_compatibility out of advanced
* remove retry_interval from UI
99% of users should never be changing this
* hide switch in optionalfieldwidget if editing a profile
* add override badges for cameras and profiles
collect shared functions into the config util and separate hooks
* Use new models endpoint info to determine modalities
* clarify language
* fix linter
---------
Co-authored-by: Nicolas Mowen <nickmowen213@gmail.com>
