remove decorator and specifically suppress TFLite delegate creation messages

* fix coral docs

* add note about sub label object classification with person

* Catch OSError for deleting classification image

* add docs for dummy camera debugging

* add to sidebar

* fix formatting

* fix

* avx instructions are required for classification

* break text on classification card to prevent button overflow

* Ensure there is no NameError when processing

* Don't use region for state classification models

* fix spelling

* Handle attribute based models

* Catch case of non-trained model that doesn't add infinite number of classification images

* Actually train object classification models automatically

---------

Co-authored-by: Nicolas Mowen <nickmowen213@gmail.com>

.github/workflows/pull_request.yml

@@ -19,9 +19,9 @@ jobs:
       - uses: actions/checkout@v6
         with:
           persist-credentials: false
-      - uses: actions/setup-node@master
+      - uses: actions/setup-node@v6
         with:
-          node-version: 16.x
+          node-version: 20.x
       - run: npm install
         working-directory: ./web
       - name: Lint
@@ -35,7 +35,7 @@ jobs:
       - uses: actions/checkout@v6
         with:
           persist-credentials: false
-      - uses: actions/setup-node@master
+      - uses: actions/setup-node@v6
         with:
           node-version: 20.x
       - run: npm install
@@ -78,7 +78,7 @@ jobs:
         uses: actions/checkout@v6
         with:
           persist-credentials: false
-      - uses: actions/setup-node@master
+      - uses: actions/setup-node@v6
         with:
           node-version: 20.x
       - name: Install devcontainer cli

README_CN.md

@@ -4,14 +4,14 @@
 
 # Frigate NVR™ - 一个具有实时目标检测的本地 NVR
 
-[English](https://github.com/blakeblackshear/frigate) | \[简体中文\]
-
-[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
-
 <a href="https://hosted.weblate.org/engage/frigate-nvr/-/zh_Hans/">
 <img src="https://hosted.weblate.org/widget/frigate-nvr/-/zh_Hans/svg-badge.svg" alt="翻译状态" />
 </a>
 
+[English](https://github.com/blakeblackshear/frigate) | \[简体中文\]
+
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+
 一个完整的本地网络视频录像机（NVR），专为[Home Assistant](https://www.home-assistant.io)设计，具备 AI 目标/物体检测功能。使用 OpenCV 和 TensorFlow 在本地为 IP 摄像头执行实时物体检测。
 
 强烈推荐使用 GPU 或者 AI 加速器（例如[Google Coral 加速器](https://coral.ai/products/) 或者 [Hailo](https://hailo.ai/)等）。它们的运行效率远远高于现在的顶级 CPU，并且功耗也极低。
@@ -38,6 +38,7 @@
 ## 协议
 
 本项目采用 **MIT 许可证**授权。
+
 **代码部分**：本代码库中的源代码、配置文件和文档均遵循 [MIT 许可证](LICENSE)。您可以自由使用、修改和分发这些代码，但必须保留原始版权声明。
 
 **商标部分**：“Frigate”名称、“Frigate NVR”品牌以及 Frigate 的 Logo 为 **Frigate LLC 的商标**，**不在** MIT 许可证覆盖范围内。

docker/main/Dockerfile

@@ -237,8 +237,18 @@ ENV PYTHONWARNINGS="ignore:::numpy.core.getlimits"
 # Set HailoRT to disable logging
 ENV HAILORT_LOGGER_PATH=NONE
 
-# TensorFlow error only
+# TensorFlow C++ logging suppression (must be set before import)
+# TF_CPP_MIN_LOG_LEVEL: 0=all, 1=INFO+, 2=WARNING+, 3=ERROR+ (we use 3 for errors only)
 ENV TF_CPP_MIN_LOG_LEVEL=3
+# Suppress verbose logging from TensorFlow C++ code
+ENV TF_CPP_MIN_VLOG_LEVEL=3
+# Disable oneDNN optimization messages ("optimized with oneDNN...")
+ENV TF_ENABLE_ONEDNN_OPTS=0
+# Suppress AutoGraph verbosity during conversion
+ENV AUTOGRAPH_VERBOSITY=0
+# Google Logging (GLOG) suppression for TensorFlow components
+ENV GLOG_minloglevel=3
+ENV GLOG_logtostderr=0
 
 ENV PATH="/usr/local/go2rtc/bin:/usr/local/tempio/bin:/usr/local/nginx/sbin:${PATH}"
 

docs/docs/configuration/authentication.md

@@ -270,3 +270,42 @@ To use role-based access control, you must connect to Frigate via the **authenti
 1. Log in as an **admin** user via port `8971`.
 2. Navigate to **Settings > Users**.
 3. Edit a user’s role by selecting **admin** or **viewer**.
+
+## API Authentication Guide
+
+### Getting a Bearer Token
+
+To use the Frigate API, you need to authenticate first. Follow these steps to obtain a Bearer token:
+
+#### 1. Login
+
+Make a POST request to `/login` with your credentials:
+
+```bash
+curl -i -X POST https://frigate_ip:8971/api/login \
+  -H "Content-Type: application/json" \
+  -d '{"user": "admin", "password": "your_password"}'
+```
+
+:::note
+
+You may need to include `-k` in the argument list in these steps (eg: `curl -k -i -X POST ...`) if your Frigate instance is using a self-signed certificate.
+
+:::
+
+The response will contain a cookie with the JWT token.
+
+#### 2. Using the Bearer Token
+
+Once you have the token, include it in the Authorization header for subsequent requests:
+
+```bash
+curl -H "Authorization: Bearer <your_token>" https://frigate_ip:8971/api/profile
+```
+
+#### 3. Token Lifecycle
+
+- Tokens are valid for the configured session length
+- Tokens are automatically refreshed when you visit the `/auth` endpoint
+- Tokens are invalidated when the user's password is changed
+- Use `/logout` to clear your session cookie

docs/docs/configuration/custom_classification/object_classification.md

@@ -11,6 +11,8 @@ Object classification models are lightweight and run very fast on CPU. Inference
 
 Training the model does briefly use a high amount of system resources for about 1–3 minutes per training run. On lower-power devices, training may take longer.
 
+A CPU with AVX instructions is required for training and inference.
+
 ## Classes
 
 Classes are the categories your model will learn to distinguish between. Each class represents a distinct visual category that the model will predict.
@@ -35,6 +37,12 @@ For object classification:
   - Ideal when multiple attributes can coexist independently.
   - Example: Detecting if a `person` in a construction yard is wearing a helmet or not.
 
+:::note
+
+A tracked object can only have a single sub label. If you are using Face Recognition and you configure an object classification model for `person` using the sub label type, your sub label may not be assigned correctly as it depends on which enrichment completes its analysis first. Consider using the `attribute` type instead.
+
+:::
+
 ## Assignment Requirements
 
 Sub labels and attributes are only assigned when both conditions are met:
@@ -94,3 +102,23 @@ When choosing which objects to classify, start with a small number of visually d
 - **Preprocessing**: Ensure examples reflect object crops similar to Frigate’s boxes; keep the subject centered.
 - **Labels**: Keep label names short and consistent; include a `none` class if you plan to ignore uncertain predictions for sub labels.
 - **Threshold**: Tune `threshold` per model to reduce false assignments. Start at `0.8` and adjust based on validation.
+
+## Debugging Classification Models
+
+To troubleshoot issues with object classification models, enable debug logging to see detailed information about classification attempts, scores, and consensus calculations.
+
+Enable debug logs for classification models by adding `frigate.data_processing.real_time.custom_classification: debug` to your `logger` configuration. These logs are verbose, so only keep this enabled when necessary. Restart Frigate after this change.
+
+```yaml
+logger:
+  default: info
+  logs:
+    frigate.data_processing.real_time.custom_classification: debug
+```
+
+The debug logs will show:
+
+- Classification probabilities for each attempt
+- Whether scores meet the threshold requirement
+- Consensus calculations and when assignments are made
+- Object classification history and weighted scores

docs/docs/configuration/custom_classification/state_classification.md

@@ -11,6 +11,8 @@ State classification models are lightweight and run very fast on CPU. Inference
 
 Training the model does briefly use a high amount of system resources for about 1–3 minutes per training run. On lower-power devices, training may take longer.
 
+A CPU with AVX instructions is required for training and inference.
+
 ## Classes
 
 Classes are the different states an area on your camera can be in. Each class represents a distinct visual state that the model will learn to recognize.
@@ -60,11 +62,9 @@ Choose one or more cameras and draw a rectangle over the area of interest for ea
 
 ### Step 3: Assign Training Examples
 
-The system will automatically generate example images from your camera feeds. You'll be guided through each class one at a time to select which images represent that state.
+The system will automatically generate example images from your camera feeds. You'll be guided through each class one at a time to select which images represent that state. It's not strictly required to select all images you see. If a state is missing from the samples, you can train it from the Recent tab later.
 
-**Important**: All images must be assigned to a state before training can begin. This includes images that may not be optimal, such as when people temporarily block the view, sun glare is present, or other distractions occur. Assign these images to the state that is actually present (based on what you know the state to be), not based on the distraction. This training helps the model correctly identify the state even when such conditions occur during inference.
-
-Once all images are assigned, training will begin automatically.
+Once some images are assigned, training will begin automatically.
 
 ### Improving the Model
 
@@ -72,3 +72,34 @@ Once all images are assigned, training will begin automatically.
 - **Data collection**: Use the model's Recent Classifications tab to gather balanced examples across times of day and weather.
 - **When to train**: Focus on cases where the model is entirely incorrect or flips between states when it should not. There's no need to train additional images when the model is already working consistently.
 - **Selecting training images**: Images scoring below 100% due to new conditions (e.g., first snow of the year, seasonal changes) or variations (e.g., objects temporarily in view, insects at night) are good candidates for training, as they represent scenarios different from the default state. Training these lower-scoring images that differ from existing training data helps prevent overfitting. Avoid training large quantities of images that look very similar, especially if they already score 100% as this can lead to overfitting.
+
+## Debugging Classification Models
+
+To troubleshoot issues with state classification models, enable debug logging to see detailed information about classification attempts, scores, and state verification.
+
+Enable debug logs for classification models by adding `frigate.data_processing.real_time.custom_classification: debug` to your `logger` configuration. These logs are verbose, so only keep this enabled when necessary. Restart Frigate after this change.
+
+```yaml
+logger:
+  default: info
+  logs:
+    frigate.data_processing.real_time.custom_classification: debug
+```
+
+The debug logs will show:
+
+- Classification probabilities for each attempt
+- Whether scores meet the threshold requirement
+- State verification progress (consecutive detections needed)
+- When state changes are published
+
+### Recent Classifications
+
+For state classification, images are only added to recent classifications under specific circumstances:
+
+- **First detection**: The first classification attempt for a camera is always saved
+- **State changes**: Images are saved when the detected state differs from the current verified state
+- **Pending verification**: Images are saved when there's a pending state change being verified (requires 3 consecutive identical states)
+- **Low confidence**: Images with scores below 100% are saved even if the state matches the current state (useful for training)
+
+Images are **not** saved when the state is stable (detected state matches current state) **and** the score is 100%. This prevents unnecessary storage of redundant high-confidence classifications.

docs/docs/configuration/hardware_acceleration_enrichments.md

@@ -13,7 +13,7 @@ Object detection and enrichments (like Semantic Search, Face Recognition, and Li
 
 - **AMD**
 
-  - ROCm will automatically be detected and used for enrichments in the `-rocm` Frigate image.
+  - ROCm support in the `-rocm` Frigate image is automatically detected for enrichments, but only some enrichment models are available due to ROCm's focus on LLMs and limited stability with certain neural network models. Frigate disables models that perform poorly or are unstable to ensure reliable operation, so only compatible enrichments may be active.
 
 - **Intel**
 

docs/docs/configuration/object_detectors.md

@@ -146,16 +146,16 @@ detectors:
 
 ### EdgeTPU Supported Models
 
-| Model                                 | Notes                                       |
-| ------------------------------------- | ------------------------------------------- |
-| [MobileNet v2](#ssdlite-mobilenet-v2) | Default model                               |
-| [YOLOv9](#yolo-v9)                    | More accurate but slower than default model |
+| Model                   | Notes                                       |
+| ----------------------- | ------------------------------------------- |
+| [Mobiledet](#mobiledet) | Default model                               |
+| [YOLOv9](#yolov9)       | More accurate but slower than default model |
 
-#### SSDLite MobileNet v2
+#### Mobiledet
 
 A TensorFlow Lite model is provided in the container at `/edgetpu_model.tflite` and is used by this detector type by default. To provide your own model, bind mount the file into the container and provide the path with `model.path`.
 
-#### YOLO v9
+#### YOLOv9
 
 [YOLOv9](https://github.com/dbro/frigate-detector-edgetpu-yolo9/releases/download/v1.0/yolov9-s-relu6-best_320_int8_edgetpu.tflite) models that are compiled for Tensorflow Lite and properly quantized are supported, but not included by default. To provide your own model, bind mount the file into the container and provide the path with `model.path`. Note that the model may require a custom label file (eg. [use this 17 label file](https://raw.githubusercontent.com/dbro/frigate-detector-edgetpu-yolo9/refs/heads/main/labels-coco17.txt) for the model linked above.)
 
@@ -175,7 +175,7 @@ model:
   width: 320 # <--- should match the imgsize of the model, typically 320
   height: 320 # <--- should match the imgsize of the model, typically 320
   path: /config/model_cache/yolov9-s-relu6-best_320_int8_edgetpu.tflite
-  labelmap_path: /labelmap/labels-coco-17.txt
+  labelmap_path: /config/labels-coco17.txt
 ```
 
 Note that the labelmap uses a subset of the complete COCO label set that has only 17 objects.

docs/docs/troubleshooting/dummy-camera.md

@@ -0,0 +1,60 @@
+---
+id: dummy-camera
+title: Troubleshooting Detection
+---
+
+When investigating object detection or tracking problems, it can be helpful to replay an exported video as a temporary "dummy" camera. This lets you reproduce issues locally, iterate on configuration (detections, zones, enrichment settings), and capture logs and clips for analysis.
+
+## When to use
+
+- Replaying an exported clip to reproduce incorrect detections
+- Testing configuration changes (model settings, trackers, filters) against a known clip
+- Gathering deterministic logs and recordings for debugging or issue reports
+
+## Example Config
+
+Place the clip you want to replay in a location accessible to Frigate (for example `/media/frigate/` or the repository `debug/` folder when developing). Then add a temporary camera to your `config/config.yml` like this:
+
+```yaml
+cameras:
+  test:
+    ffmpeg:
+      inputs:
+        - path: /media/frigate/car-stopping.mp4
+          input_args: -re -stream_loop -1 -fflags +genpts
+          roles:
+            - detect
+    detect:
+      enabled: true
+    record:
+      enabled: false
+    snapshots:
+      enabled: false
+```
+
+- `-re -stream_loop -1` tells `ffmpeg` to play the file in realtime and loop indefinitely, which is useful for long debugging sessions.
+- `-fflags +genpts` helps generate presentation timestamps when they are missing in the file.
+
+## Steps
+
+1. Export or copy the clip you want to replay to the Frigate host (e.g., `/media/frigate/` or `debug/clips/`).
+2. Add the temporary camera to `config/config.yml` (example above). Use a unique name such as `test` or `replay_camera` so it's easy to remove later.
+   - If you're debugging a specific camera, copy the settings from that camera (frame rate, model/enrichment settings, zones, etc.) into the temporary camera so the replay closely matches the original environment. Leave `record` and `snapshots` disabled unless you are specifically debugging recording or snapshot behavior.
+3. Restart Frigate.
+4. Observe the Debug view in the UI and logs as the clip is replayed. Watch detections, zones, or any feature you're looking to debug, and note any errors in the logs to reproduce the issue.
+5. Iterate on camera or enrichment settings (model, fps, zones, filters) and re-check the replay until the behavior is resolved.
+6. Remove the temporary camera from your config after debugging to avoid spurious telemetry or recordings.
+
+## Variables to consider in object tracking
+
+- The exported video will not always line up exactly with how it originally ran through Frigate (or even with the last loop). Different frames may be used on replay, which can change detections and tracking.
+- Motion detection depends on the frames used; small frame shifts can change motion regions and therefore what gets passed to the detector.
+- Object detection is not deterministic: models and post-processing can yield different results across runs, so you may not get identical detections or track IDs every time.
+
+When debugging, treat the replay as a close approximation rather than a byte-for-byte replay. Capture multiple runs, enable recording if helpful, and examine logs and saved event clips to understand variability.
+
+## Troubleshooting
+
+- No video: verify the path is correct and accessible from the Frigate process/container.
+- FFmpeg errors: check the log output for ffmpeg-specific flags and adjust `input_args` accordingly for your file/container. You may also need to disable hardware acceleration (`hwaccel_args: ""`) for the dummy camera.
+- No detections: confirm the camera `roles` include `detect`, and model/detector configuration is enabled.

docs/sidebars.ts

@@ -132,6 +132,7 @@ const sidebars: SidebarsConfig = {
       "troubleshooting/gpu",
       "troubleshooting/edgetpu",
       "troubleshooting/memory",
+      "troubleshooting/dummy-camera",
     ],
     Development: [
       "development/contributing",

docs/static/frigate-api.yaml

@@ -14,19 +14,38 @@ paths:
     get:
       tags:
         - Auth
-      summary: Auth
+      summary: Authenticate request
+      description: |-
+        Authenticates the current request based on proxy headers or JWT token.
+        This endpoint verifies authentication credentials and manages JWT token refresh.
+        On success, no JSON body is returned; authentication state is communicated via response headers and cookies.
       operationId: auth_auth_get
       responses:
-        "200":
-          description: Successful Response
-          content:
-            application/json:
-              schema: {}
+        "202":
+          description: Authentication Accepted (no response body, different headers depending on auth method)
+          headers:
+            remote-user:
+              description: Authenticated username or "viewer" in proxy-only mode
+              schema:
+                type: string
+            remote-role:
+              description: Resolved role (e.g., admin, viewer, or custom)
+              schema:
+                type: string
+            Set-Cookie:
+              description: May include refreshed JWT cookie ("frigate-token") when applicable
+              schema:
+                type: string
+        "401":
+          description: Authentication Failed
   /profile:
     get:
       tags:
         - Auth
-      summary: Profile
+      summary: Get user profile
+      description: |-
+        Returns the current authenticated user's profile including username, role, and allowed cameras.
+        This endpoint requires authentication and returns information about the user's permissions.
       operationId: profile_profile_get
       responses:
         "200":
@@ -34,11 +53,16 @@ paths:
           content:
             application/json:
               schema: {}
+        "401":
+          description: Unauthorized
   /logout:
     get:
       tags:
         - Auth
-      summary: Logout
+      summary: Logout user
+      description: |-
+        Logs out the current user by clearing the session cookie.
+        After logout, subsequent requests will require re-authentication.
       operationId: logout_logout_get
       responses:
         "200":
@@ -46,11 +70,22 @@ paths:
           content:
             application/json:
               schema: {}
+        "303":
+          description: See Other (redirects to login page)
   /login:
     post:
       tags:
         - Auth
-      summary: Login
+      summary: Login with credentials
+      description: |-
+        Authenticates a user with username and password.
+        Returns a JWT token as a secure HTTP-only cookie that can be used for subsequent API requests.
+        The JWT token can also be retrieved from the response and used as a Bearer token in the Authorization header.
+
+        Example using Bearer token:
+        ```
+        curl -H "Authorization: Bearer <token_value>" https://frigate_ip:8971/api/profile
+        ```
       operationId: login_login_post
       requestBody:
         required: true
@@ -64,6 +99,11 @@ paths:
           content:
             application/json:
               schema: {}
+        "401":
+          description: Login Failed - Invalid credentials
+          content:
+            application/json:
+              schema: {}
         "422":
           description: Validation Error
           content:
@@ -74,7 +114,10 @@ paths:
     get:
       tags:
         - Auth
-      summary: Get Users
+      summary: Get all users
+      description: |-
+        Returns a list of all users with their usernames and roles.
+        Requires admin role. Each user object contains the username and assigned role.
       operationId: get_users_users_get
       responses:
         "200":
@@ -82,10 +125,19 @@ paths:
           content:
             application/json:
               schema: {}
+        "403":
+          description: Forbidden - Admin role required
     post:
       tags:
         - Auth
-      summary: Create User
+      summary: Create new user
+      description: |-
+        Creates a new user with the specified username, password, and role.
+        Requires admin role. Password must meet strength requirements:
+        - Minimum 8 characters
+        - At least one uppercase letter
+        - At least one digit
+        - At least one special character (!@#$%^&*(),.?":{}\|<>)
       operationId: create_user_users_post
       requestBody:
         required: true
@@ -99,6 +151,13 @@ paths:
           content:
             application/json:
               schema: {}
+        "400":
+          description: Bad Request - Invalid username or role
+          content:
+            application/json:
+              schema: {}
+        "403":
+          description: Forbidden - Admin role required
         "422":
           description: Validation Error
           content:
@@ -109,7 +168,10 @@ paths:
     delete:
       tags:
         - Auth
-      summary: Delete User
+      summary: Delete user
+      description: |-
+        Deletes a user by username. The built-in admin user cannot be deleted.
+        Requires admin role. Returns success message or error if user not found.
       operationId: delete_user_users__username__delete
       parameters:
         - name: username
@@ -118,12 +180,15 @@ paths:
           schema:
             type: string
             title: Username
+            description: The username of the user to delete
       responses:
         "200":
           description: Successful Response
           content:
             application/json:
               schema: {}
+        "403":
+          description: Forbidden - Cannot delete admin user or admin role required
         "422":
           description: Validation Error
           content:
@@ -134,7 +199,17 @@ paths:
     put:
       tags:
         - Auth
-      summary: Update Password
+      summary: Update user password
+      description: |-
+        Updates a user's password. Users can only change their own password unless they have admin role.
+        Requires the current password to verify identity for non-admin users.
+        Password must meet strength requirements:
+        - Minimum 8 characters
+        - At least one uppercase letter
+        - At least one digit
+        - At least one special character (!@#$%^&*(),.?":{}\|<>)
+
+        If user changes their own password, a new JWT cookie is automatically issued.
       operationId: update_password_users__username__password_put
       parameters:
         - name: username
@@ -143,6 +218,7 @@ paths:
           schema:
             type: string
             title: Username
+            description: The username of the user whose password to update
       requestBody:
         required: true
         content:
@@ -155,6 +231,14 @@ paths:
           content:
             application/json:
               schema: {}
+        "400":
+          description: Bad Request - Current password required or password doesn't meet requirements
+        "401":
+          description: Unauthorized - Current password is incorrect
+        "403":
+          description: Forbidden - Viewers can only update their own password
+        "404":
+          description: Not Found - User not found
         "422":
           description: Validation Error
           content:
@@ -165,7 +249,10 @@ paths:
     put:
       tags:
         - Auth
-      summary: Update Role
+      summary: Update user role
+      description: |-
+        Updates a user's role. The built-in admin user's role cannot be modified.
+        Requires admin role. Valid roles are defined in the configuration.
       operationId: update_role_users__username__role_put
       parameters:
         - name: username
@@ -174,6 +261,7 @@ paths:
           schema:
             type: string
             title: Username
+            description: The username of the user whose role to update
       requestBody:
         required: true
         content:
@@ -186,6 +274,10 @@ paths:
           content:
             application/json:
               schema: {}
+        "400":
+          description: Bad Request - Invalid role
+        "403":
+          description: Forbidden - Cannot modify admin user's role or admin role required
         "422":
           description: Validation Error
           content:

frigate/api/auth.py

@@ -143,17 +143,6 @@ def require_admin_by_default():
     return admin_checker
 
 
-def _is_authenticated(request: Request) -> bool:
-    """
-    Helper to determine if a request is from an authenticated user.
-
-    Returns True if the request has a valid authenticated user (not anonymous).
-    Port 5000 internal requests are considered anonymous despite having admin role.
-    """
-    username = request.headers.get("remote-user")
-    return username is not None and username != "anonymous"
-
-
 def allow_public():
     """
     Override dependency to allow unauthenticated access to an endpoint.
@@ -173,27 +162,24 @@ def allow_public():
 
 def allow_any_authenticated():
     """
-    Override dependency to allow any authenticated user (bypass admin requirement).
+    Override dependency to allow any request that passed through the /auth endpoint.
 
     Allows:
-    - Port 5000 internal requests (have admin role despite anonymous user)
-    - Any authenticated user with a real username (not "anonymous")
+    - Port 5000 internal requests (remote-user: "anonymous", remote-role: "admin")
+    - Authenticated users with JWT tokens (remote-user: username)
+    - Unauthenticated requests when auth is disabled (remote-user: "viewer")
 
     Rejects:
-    - Port 8971 requests with anonymous user (auth disabled, no proxy auth)
+    - Requests with no remote-user header (did not pass through /auth endpoint)
 
     Example:
         @router.get("/authenticated-endpoint", dependencies=[Depends(allow_any_authenticated())])
     """
 
     async def auth_checker(request: Request):
-        # Port 5000 requests have admin role and should be allowed
-        role = request.headers.get("remote-role")
-        if role == "admin":
-            return
-
-        # Otherwise require a real authenticated user (not anonymous)
-        if not _is_authenticated(request):
+        # Ensure a remote-user has been set by the /auth endpoint
+        username = request.headers.get("remote-user")
+        if username is None:
             raise HTTPException(status_code=401, detail="Authentication required")
         return
 
@@ -549,7 +535,37 @@ def resolve_role(
 
 
 # Endpoints
-@router.get("/auth", dependencies=[Depends(allow_public())])
+@router.get(
+    "/auth",
+    dependencies=[Depends(allow_public())],
+    summary="Authenticate request",
+    description=(
+        "Authenticates the current request based on proxy headers or JWT token. "
+        "This endpoint verifies authentication credentials and manages JWT token refresh. "
+        "On success, no JSON body is returned; authentication state is communicated via response headers and cookies."
+    ),
+    status_code=202,
+    responses={
+        202: {
+            "description": "Authentication Accepted (no response body)",
+            "headers": {
+                "remote-user": {
+                    "description": 'Authenticated username or "viewer" in proxy-only mode',
+                    "schema": {"type": "string"},
+                },
+                "remote-role": {
+                    "description": "Resolved role (e.g., admin, viewer, or custom)",
+                    "schema": {"type": "string"},
+                },
+                "Set-Cookie": {
+                    "description": "May include refreshed JWT cookie when applicable",
+                    "schema": {"type": "string"},
+                },
+            },
+        },
+        401: {"description": "Authentication Failed"},
+    },
+)
 def auth(request: Request):
     auth_config: AuthConfig = request.app.frigate_config.auth
     proxy_config: ProxyConfig = request.app.frigate_config.proxy
@@ -576,12 +592,12 @@ def auth(request: Request):
     # if auth is disabled, just apply the proxy header map and return success
     if not auth_config.enabled:
         # pass the user header value from the upstream proxy if a mapping is specified
-        # or use anonymous if none are specified
+        # or use viewer if none are specified
         user_header = proxy_config.header_map.user
         success_response.headers["remote-user"] = (
-            request.headers.get(user_header, default="anonymous")
+            request.headers.get(user_header, default="viewer")
             if user_header
-            else "anonymous"
+            else "viewer"
         )
 
         # parse header and resolve a valid role
@@ -689,9 +705,14 @@ def auth(request: Request):
         return fail_response
 
 
-@router.get("/profile", dependencies=[Depends(allow_any_authenticated())])
+@router.get(
+    "/profile",
+    dependencies=[Depends(allow_any_authenticated())],
+    summary="Get user profile",
+    description="Returns the current authenticated user's profile including username, role, and allowed cameras. This endpoint requires authentication and returns information about the user's permissions.",
+)
 def profile(request: Request):
-    username = request.headers.get("remote-user", "anonymous")
+    username = request.headers.get("remote-user", "viewer")
     role = request.headers.get("remote-role", "viewer")
 
     all_camera_names = set(request.app.frigate_config.cameras.keys())
@@ -703,7 +724,12 @@ def profile(request: Request):
     )
 
 
-@router.get("/logout", dependencies=[Depends(allow_public())])
+@router.get(
+    "/logout",
+    dependencies=[Depends(allow_public())],
+    summary="Logout user",
+    description="Logs out the current user by clearing the session cookie. After logout, subsequent requests will require re-authentication.",
+)
 def logout(request: Request):
     auth_config: AuthConfig = request.app.frigate_config.auth
     response = RedirectResponse("/login", status_code=303)
@@ -714,7 +740,12 @@ def logout(request: Request):
 limiter = Limiter(key_func=get_remote_addr)
 
 
-@router.post("/login", dependencies=[Depends(allow_public())])
+@router.post(
+    "/login",
+    dependencies=[Depends(allow_public())],
+    summary="Login with credentials",
+    description='Authenticates a user with username and password. Returns a JWT token as a secure HTTP-only cookie that can be used for subsequent API requests. The JWT token can also be retrieved from the response and used as a Bearer token in the Authorization header.\n\nExample using Bearer token:\n```\ncurl -H "Authorization: Bearer <token_value>" https://frigate_ip:8971/api/profile\n```',
+)
 @limiter.limit(limit_value=rateLimiter.get_limit)
 def login(request: Request, body: AppPostLoginBody):
     JWT_COOKIE_NAME = request.app.frigate_config.auth.cookie_name
@@ -752,7 +783,12 @@ def login(request: Request, body: AppPostLoginBody):
     return JSONResponse(content={"message": "Login failed"}, status_code=401)
 
 
-@router.get("/users", dependencies=[Depends(require_role(["admin"]))])
+@router.get(
+    "/users",
+    dependencies=[Depends(require_role(["admin"]))],
+    summary="Get all users",
+    description="Returns a list of all users with their usernames and roles. Requires admin role. Each user object contains the username and assigned role.",
+)
 def get_users():
     exports = (
         User.select(User.username, User.role).order_by(User.username).dicts().iterator()
@@ -760,7 +796,12 @@ def get_users():
     return JSONResponse([e for e in exports])
 
 
-@router.post("/users", dependencies=[Depends(require_role(["admin"]))])
+@router.post(
+    "/users",
+    dependencies=[Depends(require_role(["admin"]))],
+    summary="Create new user",
+    description='Creates a new user with the specified username, password, and role. Requires admin role. Password must meet strength requirements: minimum 8 characters, at least one uppercase letter, at least one digit, and at least one special character (!@#$%^&*(),.?":{} |<>).',
+)
 def create_user(
     request: Request,
     body: AppPostUsersBody,
@@ -789,7 +830,12 @@ def create_user(
     return JSONResponse(content={"username": body.username})
 
 
-@router.delete("/users/{username}", dependencies=[Depends(require_role(["admin"]))])
+@router.delete(
+    "/users/{username}",
+    dependencies=[Depends(require_role(["admin"]))],
+    summary="Delete user",
+    description="Deletes a user by username. The built-in admin user cannot be deleted. Requires admin role. Returns success message or error if user not found.",
+)
 def delete_user(request: Request, username: str):
     # Prevent deletion of the built-in admin user
     if username == "admin":
@@ -802,7 +848,10 @@ def delete_user(request: Request, username: str):
 
 
 @router.put(
-    "/users/{username}/password", dependencies=[Depends(allow_any_authenticated())]
+    "/users/{username}/password",
+    dependencies=[Depends(allow_any_authenticated())],
+    summary="Update user password",
+    description="Updates a user's password. Users can only change their own password unless they have admin role. Requires the current password to verify identity for non-admin users. Password must meet strength requirements: minimum 8 characters, at least one uppercase letter, at least one digit, and at least one special character (!@#$%^&*(),.?\":{} |<>). If user changes their own password, a new JWT cookie is automatically issued.",
 )
 async def update_password(
     request: Request,
@@ -830,13 +879,9 @@ async def update_password(
     except DoesNotExist:
         return JSONResponse(content={"message": "User not found"}, status_code=404)
 
-    # Require old_password when:
-    # 1. Non-admin user is changing another user's password (admin only action)
-    # 2. Any user is changing their own password
-    is_changing_own_password = current_username == username
-    is_non_admin = current_role != "admin"
-
-    if is_changing_own_password or is_non_admin:
+    # Require old_password when non-admin user is changing any password
+    # Admin users changing passwords do NOT need to provide the current password
+    if current_role != "admin":
         if not body.old_password:
             return JSONResponse(
                 content={"message": "Current password is required"},
@@ -887,6 +932,8 @@ async def update_password(
 @router.put(
     "/users/{username}/role",
     dependencies=[Depends(require_role(["admin"]))],
+    summary="Update user role",
+    description="Updates a user's role. The built-in admin user's role cannot be modified. Requires admin role. Valid roles are defined in the configuration.",
 )
 async def update_role(
     request: Request,

frigate/comms/mqtt.py

@@ -225,7 +225,8 @@ class MqttClient(Communicator):
             "birdseye_mode",
             "review_alerts",
             "review_detections",
-            "genai",
+            "object_descriptions",
+            "review_descriptions",
         ]
 
         for name in self.config.cameras.keys():

frigate/const.py

@@ -77,6 +77,9 @@ FFMPEG_HWACCEL_RKMPP = "preset-rkmpp"
 FFMPEG_HWACCEL_AMF = "preset-amd-amf"
 FFMPEG_HVC1_ARGS = ["-tag:v", "hvc1"]
 
+# RKNN constants
+SUPPORTED_RK_SOCS = ["rk3562", "rk3566", "rk3568", "rk3576", "rk3588"]
+
 # Regex constants
 
 REGEX_CAMERA_NAME = r"^[a-zA-Z0-9_-]+$"

frigate/data_processing/real_time/bird.py

@@ -13,7 +13,7 @@ from frigate.comms.event_metadata_updater import (
 )
 from frigate.config import FrigateConfig
 from frigate.const import MODEL_CACHE_DIR
-from frigate.log import redirect_output_to_logger
+from frigate.log import suppress_stderr_during
 from frigate.util.object import calculate_region
 
 from ..types import DataProcessorMetrics
@@ -80,13 +80,14 @@ class BirdRealTimeProcessor(RealTimeProcessorApi):
         except Exception as e:
             logger.error(f"Failed to download {path}: {e}")
 
-    @redirect_output_to_logger(logger, logging.DEBUG)
     def __build_detector(self) -> None:
-        self.interpreter = Interpreter(
-            model_path=os.path.join(MODEL_CACHE_DIR, "bird/bird.tflite"),
-            num_threads=2,
-        )
-        self.interpreter.allocate_tensors()
+        # Suppress TFLite delegate creation messages that bypass Python logging
+        with suppress_stderr_during("tflite_interpreter_init"):
+            self.interpreter = Interpreter(
+                model_path=os.path.join(MODEL_CACHE_DIR, "bird/bird.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()
 

frigate/data_processing/real_time/custom_classification.py

@@ -21,7 +21,7 @@ from frigate.config.classification import (
     ObjectClassificationType,
 )
 from frigate.const import CLIPS_DIR, MODEL_CACHE_DIR
-from frigate.log import redirect_output_to_logger
+from frigate.log import suppress_stderr_during
 from frigate.types import TrackedObjectUpdateTypesEnum
 from frigate.util.builtin import EventsPerSecond, InferenceSpeed, load_labels
 from frigate.util.object import box_overlaps, calculate_region
@@ -52,7 +52,7 @@ class CustomStateClassificationProcessor(RealTimeProcessorApi):
         self.requestor = requestor
         self.model_dir = os.path.join(MODEL_CACHE_DIR, self.model_config.name)
         self.train_dir = os.path.join(CLIPS_DIR, self.model_config.name, "train")
-        self.interpreter: Interpreter | None = None
+        self.interpreter: Interpreter = None
         self.tensor_input_details: dict[str, Any] | None = None
         self.tensor_output_details: dict[str, Any] | None = None
         self.labelmap: dict[int, str] = {}
@@ -72,8 +72,12 @@ class CustomStateClassificationProcessor(RealTimeProcessorApi):
         self.last_run = datetime.datetime.now().timestamp()
         self.__build_detector()
 
-    @redirect_output_to_logger(logger, logging.DEBUG)
     def __build_detector(self) -> None:
+        try:
+            from tflite_runtime.interpreter import Interpreter
+        except ModuleNotFoundError:
+            from tensorflow.lite.python.interpreter import Interpreter
+
         model_path = os.path.join(self.model_dir, "model.tflite")
         labelmap_path = os.path.join(self.model_dir, "labelmap.txt")
 
@@ -84,11 +88,13 @@ class CustomStateClassificationProcessor(RealTimeProcessorApi):
             self.labelmap = {}
             return
 
-        self.interpreter = Interpreter(
-            model_path=model_path,
-            num_threads=2,
-        )
-        self.interpreter.allocate_tensors()
+        # Suppress TFLite delegate creation messages that bypass Python logging
+        with suppress_stderr_during("tflite_interpreter_init"):
+            self.interpreter = Interpreter(
+                model_path=model_path,
+                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()
         self.labelmap = load_labels(labelmap_path, prefill=0)
@@ -224,28 +230,34 @@ class CustomStateClassificationProcessor(RealTimeProcessorApi):
         if not should_run:
             return
 
-        x, y, x2, y2 = calculate_region(
-            frame.shape,
-            crop[0],
-            crop[1],
-            crop[2],
-            crop[3],
-            224,
-            1.0,
-        )
-
         rgb = cv2.cvtColor(frame, cv2.COLOR_YUV2RGB_I420)
-        frame = rgb[
-            y:y2,
-            x:x2,
-        ]
+        height, width = rgb.shape[:2]
 
-        if frame.shape != (224, 224):
-            try:
-                resized_frame = cv2.resize(frame, (224, 224))
-            except Exception:
-                logger.warning("Failed to resize image for state classification")
-                return
+        # Convert normalized crop coordinates to pixel values
+        x1 = int(camera_config.crop[0] * width)
+        y1 = int(camera_config.crop[1] * height)
+        x2 = int(camera_config.crop[2] * width)
+        y2 = int(camera_config.crop[3] * height)
+
+        # Clip coordinates to frame boundaries
+        x1 = max(0, min(x1, width))
+        y1 = max(0, min(y1, height))
+        x2 = max(0, min(x2, width))
+        y2 = max(0, min(y2, height))
+
+        if x2 <= x1 or y2 <= y1:
+            logger.warning(
+                f"Invalid crop coordinates for {camera}: [{x1}, {y1}, {x2}, {y2}]"
+            )
+            return
+
+        frame = rgb[y1:y2, x1:x2]
+
+        try:
+            resized_frame = cv2.resize(frame, (224, 224))
+        except Exception:
+            logger.warning("Failed to resize image for state classification")
+            return
 
         if self.interpreter is None:
             # When interpreter is None, always save (score is 0.0, which is < 1.0)
@@ -345,7 +357,7 @@ class CustomObjectClassificationProcessor(RealTimeProcessorApi):
         self.model_config = model_config
         self.model_dir = os.path.join(MODEL_CACHE_DIR, self.model_config.name)
         self.train_dir = os.path.join(CLIPS_DIR, self.model_config.name, "train")
-        self.interpreter: Interpreter | None = None
+        self.interpreter: Interpreter = None
         self.sub_label_publisher = sub_label_publisher
         self.requestor = requestor
         self.tensor_input_details: dict[str, Any] | None = None
@@ -366,7 +378,6 @@ class CustomObjectClassificationProcessor(RealTimeProcessorApi):
 
         self.__build_detector()
 
-    @redirect_output_to_logger(logger, logging.DEBUG)
     def __build_detector(self) -> None:
         model_path = os.path.join(self.model_dir, "model.tflite")
         labelmap_path = os.path.join(self.model_dir, "labelmap.txt")
@@ -378,11 +389,13 @@ class CustomObjectClassificationProcessor(RealTimeProcessorApi):
             self.labelmap = {}
             return
 
-        self.interpreter = Interpreter(
-            model_path=model_path,
-            num_threads=2,
-        )
-        self.interpreter.allocate_tensors()
+        # Suppress TFLite delegate creation messages that bypass Python logging
+        with suppress_stderr_during("tflite_interpreter_init"):
+            self.interpreter = Interpreter(
+                model_path=model_path,
+                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()
         self.labelmap = load_labels(labelmap_path, prefill=0)
@@ -508,6 +521,13 @@ class CustomObjectClassificationProcessor(RealTimeProcessorApi):
                 0.0,
                 max_files=save_attempts,
             )
+
+            # Still track history even when model doesn't exist to respect MAX_OBJECT_CLASSIFICATIONS
+            # Add an entry with "unknown" label so the history limit is enforced
+            if object_id not in self.classification_history:
+                self.classification_history[object_id] = []
+
+            self.classification_history[object_id].append(("unknown", 0.0, now))
             return
 
         input = np.expand_dims(resized_crop, axis=0)
@@ -649,5 +669,5 @@ def write_classification_attempt(
 
         if len(files) > max_files:
             os.unlink(os.path.join(folder, files[-1]))
-    except FileNotFoundError:
+    except (FileNotFoundError, OSError):
         pass

frigate/detectors/detection_runners.py

@@ -131,6 +131,7 @@ class ONNXModelRunner(BaseModelRunner):
 
         return model_type in [
             EnrichmentModelTypeEnum.paddleocr.value,
+            EnrichmentModelTypeEnum.yolov9_license_plate.value,
             EnrichmentModelTypeEnum.jina_v1.value,
             EnrichmentModelTypeEnum.jina_v2.value,
             EnrichmentModelTypeEnum.facenet.value,
@@ -169,6 +170,7 @@ class CudaGraphRunner(BaseModelRunner):
 
         return model_type not in [
             ModelTypeEnum.yolonas.value,
+            ModelTypeEnum.dfine.value,
             EnrichmentModelTypeEnum.paddleocr.value,
             EnrichmentModelTypeEnum.jina_v1.value,
             EnrichmentModelTypeEnum.jina_v2.value,

frigate/detectors/plugins/cpu_tfl.py

@@ -5,7 +5,7 @@ from typing_extensions import Literal
 
 from frigate.detectors.detection_api import DetectionApi
 from frigate.detectors.detector_config import BaseDetectorConfig
-from frigate.log import redirect_output_to_logger
+from frigate.log import suppress_stderr_during
 
 from ..detector_utils import tflite_detect_raw, tflite_init
 
@@ -28,12 +28,13 @@ class CpuDetectorConfig(BaseDetectorConfig):
 class CpuTfl(DetectionApi):
     type_key = DETECTOR_KEY
 
-    @redirect_output_to_logger(logger, logging.DEBUG)
     def __init__(self, detector_config: CpuDetectorConfig):
-        interpreter = Interpreter(
-            model_path=detector_config.model.path,
-            num_threads=detector_config.num_threads or 3,
-        )
+        # Suppress TFLite delegate creation messages that bypass Python logging
+        with suppress_stderr_during("tflite_interpreter_init"):
+            interpreter = Interpreter(
+                model_path=detector_config.model.path,
+                num_threads=detector_config.num_threads or 3,
+            )
 
         tflite_init(self, interpreter)
 

frigate/detectors/plugins/rknn.py

@@ -8,7 +8,7 @@ import cv2
 import numpy as np
 from pydantic import Field
 
-from frigate.const import MODEL_CACHE_DIR
+from frigate.const import MODEL_CACHE_DIR, SUPPORTED_RK_SOCS
 from frigate.detectors.detection_api import DetectionApi
 from frigate.detectors.detection_runners import RKNNModelRunner
 from frigate.detectors.detector_config import BaseDetectorConfig, ModelTypeEnum
@@ -19,8 +19,6 @@ logger = logging.getLogger(__name__)
 
 DETECTOR_KEY = "rknn"
 
-supported_socs = ["rk3562", "rk3566", "rk3568", "rk3576", "rk3588"]
-
 supported_models = {
     ModelTypeEnum.yologeneric: "^frigate-fp16-yolov9-[cemst]$",
     ModelTypeEnum.yolonas: "^deci-fp16-yolonas_[sml]$",
@@ -82,9 +80,9 @@ class Rknn(DetectionApi):
         except FileNotFoundError:
             raise Exception("Make sure to run docker in privileged mode.")
 
-        if soc not in supported_socs:
+        if soc not in SUPPORTED_RK_SOCS:
             raise Exception(
-                f"Your SoC is not supported. Your SoC is: {soc}. Currently these SoCs are supported: {supported_socs}."
+                f"Your SoC is not supported. Your SoC is: {soc}. Currently these SoCs are supported: {SUPPORTED_RK_SOCS}."
             )
 
         return soc

frigate/embeddings/onnx/face_embedding.py

@@ -8,7 +8,7 @@ import numpy as np
 from frigate.const import MODEL_CACHE_DIR
 from frigate.detectors.detection_runners import get_optimized_runner
 from frigate.embeddings.types import EnrichmentModelTypeEnum
-from frigate.log import redirect_output_to_logger
+from frigate.log import suppress_stderr_during
 from frigate.util.downloader import ModelDownloader
 
 from ...config import FaceRecognitionConfig
@@ -57,17 +57,18 @@ class FaceNetEmbedding(BaseEmbedding):
             self._load_model_and_utils()
             logger.debug(f"models are already downloaded for {self.model_name}")
 
-    @redirect_output_to_logger(logger, logging.DEBUG)
     def _load_model_and_utils(self):
         if self.runner is None:
             if self.downloader:
                 self.downloader.wait_for_download()
 
-            self.runner = Interpreter(
-                model_path=os.path.join(MODEL_CACHE_DIR, "facedet/facenet.tflite"),
-                num_threads=2,
-            )
-            self.runner.allocate_tensors()
+            # Suppress TFLite delegate creation messages that bypass Python logging
+            with suppress_stderr_during("tflite_interpreter_init"):
+                self.runner = Interpreter(
+                    model_path=os.path.join(MODEL_CACHE_DIR, "facedet/facenet.tflite"),
+                    num_threads=2,
+                )
+                self.runner.allocate_tensors()
             self.tensor_input_details = self.runner.get_input_details()
             self.tensor_output_details = self.runner.get_output_details()
 

frigate/embeddings/onnx/jina_v1_embedding.py

@@ -186,6 +186,9 @@ class JinaV1ImageEmbedding(BaseEmbedding):
                 download_func=self._download_model,
             )
             self.downloader.ensure_model_files()
+            # Avoid lazy loading in worker threads: block until downloads complete
+            # and load the model on the main thread during initialization.
+            self._load_model_and_utils()
         else:
             self.downloader = None
             ModelDownloader.mark_files_state(

frigate/embeddings/onnx/jina_v2_embedding.py

@@ -65,6 +65,9 @@ class JinaV2Embedding(BaseEmbedding):
                 download_func=self._download_model,
             )
             self.downloader.ensure_model_files()
+            # Avoid lazy loading in worker threads: block until downloads complete
+            # and load the model on the main thread during initialization.
+            self._load_model_and_utils()
         else:
             self.downloader = None
             ModelDownloader.mark_files_state(

frigate/events/audio.py

@@ -34,7 +34,7 @@ from frigate.data_processing.real_time.audio_transcription import (
     AudioTranscriptionRealTimeProcessor,
 )
 from frigate.ffmpeg_presets import parse_preset_input
-from frigate.log import LogPipe, redirect_output_to_logger
+from frigate.log import LogPipe, suppress_stderr_during
 from frigate.object_detection.base import load_labels
 from frigate.util.builtin import get_ffmpeg_arg_list
 from frigate.util.process import FrigateProcess
@@ -367,17 +367,17 @@ class AudioEventMaintainer(threading.Thread):
 
 
 class AudioTfl:
-    @redirect_output_to_logger(logger, logging.DEBUG)
     def __init__(self, stop_event: threading.Event, num_threads=2):
         self.stop_event = stop_event
         self.num_threads = num_threads
         self.labels = load_labels("/audio-labelmap.txt", prefill=521)
-        self.interpreter = Interpreter(
-            model_path="/cpu_audio_model.tflite",
-            num_threads=self.num_threads,
-        )
-
-        self.interpreter.allocate_tensors()
+        # Suppress TFLite delegate creation messages that bypass Python logging
+        with suppress_stderr_during("tflite_interpreter_init"):
+            self.interpreter = Interpreter(
+                model_path="/cpu_audio_model.tflite",
+                num_threads=self.num_threads,
+            )
+            self.interpreter.allocate_tensors()
 
         self.tensor_input_details = self.interpreter.get_input_details()
         self.tensor_output_details = self.interpreter.get_output_details()

frigate/ffmpeg_presets.py

@@ -153,7 +153,7 @@ PRESETS_HW_ACCEL_ENCODE_BIRDSEYE = {
     FFMPEG_HWACCEL_VAAPI: "{0} -hide_banner -hwaccel vaapi -hwaccel_output_format vaapi -hwaccel_device {3} {1} -c:v h264_vaapi -g 50 -bf 0 -profile:v high -level:v 4.1 -sei:v 0 -an -vf format=vaapi|nv12,hwupload {2}",
     "preset-intel-qsv-h264": "{0} -hide_banner {1} -c:v h264_qsv -g 50 -bf 0 -profile:v high -level:v 4.1 -async_depth:v 1 {2}",
     "preset-intel-qsv-h265": "{0} -hide_banner {1} -c:v h264_qsv -g 50 -bf 0 -profile:v main -level:v 4.1 -async_depth:v 1 {2}",
-    FFMPEG_HWACCEL_NVIDIA: "{0} -hide_banner {1} -hwaccel device {3} -c:v h264_nvenc -g 50 -profile:v high -level:v auto -preset:v p2 -tune:v ll {2}",
+    FFMPEG_HWACCEL_NVIDIA: "{0} -hide_banner {1} -c:v h264_nvenc -g 50 -profile:v high -level:v auto -preset:v p2 -tune:v ll {2}",
     "preset-jetson-h264": "{0} -hide_banner {1} -c:v h264_nvmpi -profile high {2}",
     "preset-jetson-h265": "{0} -hide_banner {1} -c:v h264_nvmpi -profile main {2}",
     FFMPEG_HWACCEL_RKMPP: "{0} -hide_banner {1} -c:v h264_rkmpp -profile:v high {2}",

frigate/log.py

@@ -80,10 +80,15 @@ def apply_log_levels(default: str, log_levels: dict[str, LogLevel]) -> None:
     log_levels = {
         "absl": LogLevel.error,
         "httpx": LogLevel.error,
+        "h5py": LogLevel.error,
+        "keras": LogLevel.error,
         "matplotlib": LogLevel.error,
         "tensorflow": LogLevel.error,
+        "tensorflow.python": LogLevel.error,
         "werkzeug": LogLevel.error,
         "ws4py": LogLevel.error,
+        "PIL": LogLevel.warning,
+        "numba": LogLevel.warning,
         **log_levels,
     }
 
@@ -318,3 +323,31 @@ def suppress_os_output(func: Callable) -> Callable:
         return result
 
     return wrapper
+
+
+@contextmanager
+def suppress_stderr_during(operation_name: str) -> Generator[None, None, None]:
+    """
+    Context manager to suppress stderr output during a specific operation.
+
+    Useful for silencing LLVM debug output, CUDA messages, and other native
+    library logging that cannot be controlled via Python logging or environment
+    variables. Completely redirects file descriptor 2 (stderr) to /dev/null.
+
+    Usage:
+        with suppress_stderr_during("model_conversion"):
+            converter = tf.lite.TFLiteConverter.from_keras_model(model)
+            tflite_model = converter.convert()
+
+    Args:
+        operation_name: Name of the operation for debugging purposes
+    """
+    original_stderr_fd = os.dup(2)
+    devnull = os.open(os.devnull, os.O_WRONLY)
+    try:
+        os.dup2(devnull, 2)
+        yield
+    finally:
+        os.dup2(original_stderr_fd, 2)
+        os.close(devnull)
+        os.close(original_stderr_fd)

frigate/record/cleanup.py

@@ -119,6 +119,7 @@ class RecordingCleanup(threading.Thread):
                 Recordings.path,
                 Recordings.objects,
                 Recordings.motion,
+                Recordings.dBFS,
             )
             .where(
                 (Recordings.camera == config.name)
@@ -126,6 +127,7 @@ class RecordingCleanup(threading.Thread):
                     (
                         (Recordings.end_time < continuous_expire_date)
                         & (Recordings.motion == 0)
+                        & (Recordings.dBFS == 0)
                     )
                     | (Recordings.end_time < motion_expire_date)
                 )
@@ -185,6 +187,7 @@ class RecordingCleanup(threading.Thread):
                     mode == RetainModeEnum.motion
                     and recording.motion == 0
                     and recording.objects == 0
+                    and recording.dBFS == 0
                 )
                 or (mode == RetainModeEnum.active_objects and recording.objects == 0)
             ):

frigate/record/maintainer.py

@@ -67,7 +67,7 @@ class SegmentInfo:
         if (
             not keep
             and retain_mode == RetainModeEnum.motion
-            and (self.motion_count > 0 or self.average_dBFS > 0)
+            and (self.motion_count > 0 or self.average_dBFS != 0)
         ):
             keep = True
 

frigate/util/classification.py

@@ -19,9 +19,10 @@ from frigate.const import (
     PROCESS_PRIORITY_LOW,
     UPDATE_MODEL_STATE,
 )
-from frigate.log import redirect_output_to_logger
+from frigate.log import redirect_output_to_logger, suppress_stderr_during
 from frigate.models import Event, Recordings, ReviewSegment
 from frigate.types import ModelStatusTypesEnum
+from frigate.util.downloader import ModelDownloader
 from frigate.util.file import get_event_thumbnail_bytes
 from frigate.util.image import get_image_from_recording
 from frigate.util.process import FrigateProcess
@@ -121,6 +122,10 @@ def get_dataset_image_count(model_name: str) -> int:
 
 class ClassificationTrainingProcess(FrigateProcess):
     def __init__(self, model_name: str) -> None:
+        self.BASE_WEIGHT_URL = os.environ.get(
+            "TF_KERAS_MOBILENET_V2_WEIGHTS_URL",
+            "",
+        )
         super().__init__(
             stop_event=None,
             priority=PROCESS_PRIORITY_LOW,
@@ -179,11 +184,23 @@ class ClassificationTrainingProcess(FrigateProcess):
                 )
                 return False
 
+            weights_path = "imagenet"
+            # Download MobileNetV2 weights if not present
+            if self.BASE_WEIGHT_URL:
+                weights_path = os.path.join(
+                    MODEL_CACHE_DIR, "MobileNet", "mobilenet_v2_weights.h5"
+                )
+                if not os.path.exists(weights_path):
+                    logger.info("Downloading MobileNet V2 weights file")
+                    ModelDownloader.download_from_url(
+                        self.BASE_WEIGHT_URL, weights_path
+                    )
+
             # Start with imagenet base model with 35% of channels in each layer
             base_model = MobileNetV2(
                 input_shape=(224, 224, 3),
                 include_top=False,
-                weights="imagenet",
+                weights=weights_path,
                 alpha=0.35,
             )
             base_model.trainable = False  # Freeze pre-trained layers
@@ -233,15 +250,20 @@ class ClassificationTrainingProcess(FrigateProcess):
             logger.debug(f"Converting {self.model_name} to TFLite...")
 
             # convert model to tflite
-            converter = tf.lite.TFLiteConverter.from_keras_model(model)
-            converter.optimizations = [tf.lite.Optimize.DEFAULT]
-            converter.representative_dataset = (
-                self.__generate_representative_dataset_factory(dataset_dir)
-            )
-            converter.target_spec.supported_ops = [tf.lite.OpsSet.TFLITE_BUILTINS_INT8]
-            converter.inference_input_type = tf.uint8
-            converter.inference_output_type = tf.uint8
-            tflite_model = converter.convert()
+            # Suppress stderr during conversion to avoid LLVM debug output
+            # (fully_quantize, inference_type, MLIR optimization messages, etc)
+            with suppress_stderr_during("tflite_conversion"):
+                converter = tf.lite.TFLiteConverter.from_keras_model(model)
+                converter.optimizations = [tf.lite.Optimize.DEFAULT]
+                converter.representative_dataset = (
+                    self.__generate_representative_dataset_factory(dataset_dir)
+                )
+                converter.target_spec.supported_ops = [
+                    tf.lite.OpsSet.TFLITE_BUILTINS_INT8
+                ]
+                converter.inference_input_type = tf.uint8
+                converter.inference_output_type = tf.uint8
+                tflite_model = converter.convert()
 
             # write model
             model_path = os.path.join(model_dir, "model.tflite")
@@ -338,8 +360,6 @@ def collect_state_classification_examples(
         cameras: Dict mapping camera names to normalized crop coordinates [x1, y1, x2, y2] (0-1)
     """
     dataset_dir = os.path.join(CLIPS_DIR, model_name, "dataset")
-    temp_dir = os.path.join(dataset_dir, "temp")
-    os.makedirs(temp_dir, exist_ok=True)
 
     # Step 1: Get review items for the cameras
     camera_names = list(cameras.keys())
@@ -354,6 +374,10 @@ def collect_state_classification_examples(
         logger.warning(f"No review items found for cameras: {camera_names}")
         return
 
+    # The temp directory is only created when there are review_items.
+    temp_dir = os.path.join(dataset_dir, "temp")
+    os.makedirs(temp_dir, exist_ok=True)
+
     # Step 2: Create balanced timestamp selection (100 samples)
     timestamps = _select_balanced_timestamps(review_items, target_count=100)
 
@@ -482,6 +506,10 @@ def _extract_keyframes(
     """
     Extract keyframes from recordings at specified timestamps and crop to specified regions.
 
+    This implementation batches work by running multiple ffmpeg snapshot commands
+    concurrently, which significantly reduces total runtime compared to
+    processing each timestamp serially.
+
     Args:
         ffmpeg_path: Path to ffmpeg binary
         timestamps: List of timestamp dicts from _select_balanced_timestamps
@@ -491,15 +519,21 @@ def _extract_keyframes(
     Returns:
         List of paths to successfully extracted and cropped keyframe images
     """
-    keyframe_paths = []
+    from concurrent.futures import ThreadPoolExecutor, as_completed
 
-    for idx, ts_info in enumerate(timestamps):
+    if not timestamps:
+        return []
+
+    # Limit the number of concurrent ffmpeg processes so we don't overload the host.
+    max_workers = min(5, len(timestamps))
+
+    def _process_timestamp(idx: int, ts_info: dict) -> tuple[int, str | None]:
         camera = ts_info["camera"]
         timestamp = ts_info["timestamp"]
 
         if camera not in camera_crops:
             logger.warning(f"No crop coordinates for camera {camera}")
-            continue
+            return idx, None
 
         norm_x1, norm_y1, norm_x2, norm_y2 = camera_crops[camera]
 
@@ -516,7 +550,7 @@ def _extract_keyframes(
                 .get()
             )
         except Exception:
-            continue
+            return idx, None
 
         relative_time = timestamp - recording.start_time
 
@@ -530,38 +564,57 @@ def _extract_keyframes(
                 height=None,
             )
 
-            if image_data:
-                nparr = np.frombuffer(image_data, np.uint8)
-                img = cv2.imdecode(nparr, cv2.IMREAD_COLOR)
+            if not image_data:
+                return idx, None
 
-                if img is not None:
-                    height, width = img.shape[:2]
+            nparr = np.frombuffer(image_data, np.uint8)
+            img = cv2.imdecode(nparr, cv2.IMREAD_COLOR)
 
-                    x1 = int(norm_x1 * width)
-                    y1 = int(norm_y1 * height)
-                    x2 = int(norm_x2 * width)
-                    y2 = int(norm_y2 * height)
+            if img is None:
+                return idx, None
 
-                    x1_clipped = max(0, min(x1, width))
-                    y1_clipped = max(0, min(y1, height))
-                    x2_clipped = max(0, min(x2, width))
-                    y2_clipped = max(0, min(y2, height))
+            height, width = img.shape[:2]
 
-                    if x2_clipped > x1_clipped and y2_clipped > y1_clipped:
-                        cropped = img[y1_clipped:y2_clipped, x1_clipped:x2_clipped]
-                        resized = cv2.resize(cropped, (224, 224))
+            x1 = int(norm_x1 * width)
+            y1 = int(norm_y1 * height)
+            x2 = int(norm_x2 * width)
+            y2 = int(norm_y2 * height)
 
-                        output_path = os.path.join(output_dir, f"frame_{idx:04d}.jpg")
-                        cv2.imwrite(output_path, resized)
-                        keyframe_paths.append(output_path)
+            x1_clipped = max(0, min(x1, width))
+            y1_clipped = max(0, min(y1, height))
+            x2_clipped = max(0, min(x2, width))
+            y2_clipped = max(0, min(y2, height))
 
+            if x2_clipped <= x1_clipped or y2_clipped <= y1_clipped:
+                return idx, None
+
+            cropped = img[y1_clipped:y2_clipped, x1_clipped:x2_clipped]
+            resized = cv2.resize(cropped, (224, 224))
+
+            output_path = os.path.join(output_dir, f"frame_{idx:04d}.jpg")
+            cv2.imwrite(output_path, resized)
+            return idx, output_path
         except Exception as e:
             logger.debug(
                 f"Failed to extract frame from {recording.path} at {relative_time}s: {e}"
             )
-            continue
+            return idx, None
 
-    return keyframe_paths
+    keyframes_with_index: list[tuple[int, str]] = []
+
+    with ThreadPoolExecutor(max_workers=max_workers) as executor:
+        future_to_idx = {
+            executor.submit(_process_timestamp, idx, ts_info): idx
+            for idx, ts_info in enumerate(timestamps)
+        }
+
+        for future in as_completed(future_to_idx):
+            _, path = future.result()
+            if path:
+                keyframes_with_index.append((future_to_idx[future], path))
+
+    keyframes_with_index.sort(key=lambda item: item[0])
+    return [path for _, path in keyframes_with_index]
 
 
 def _select_distinct_images(

frigate/util/process.py

@@ -65,10 +65,15 @@ class FrigateProcess(BaseProcess):
         logging.basicConfig(handlers=[], force=True)
         logging.getLogger().addHandler(QueueHandler(self.__log_queue))
 
+        # Always apply base log level suppressions for noisy third-party libraries
+        # even if no specific logConfig is provided
         if logConfig:
             frigate.log.apply_log_levels(
                 logConfig.default.value.upper(), logConfig.logs
             )
+        else:
+            # Apply default INFO level with standard library suppressions
+            frigate.log.apply_log_levels("INFO", {})
 
         self._setup_memray()
 

frigate/util/rknn_converter.py

@@ -8,6 +8,7 @@ import time
 from pathlib import Path
 from typing import Optional
 
+from frigate.const import SUPPORTED_RK_SOCS
 from frigate.util.file import FileLock
 
 logger = logging.getLogger(__name__)
@@ -68,9 +69,20 @@ def is_rknn_compatible(model_path: str, model_type: str | None = None) -> bool:
         True if the model is RKNN-compatible, False otherwise
     """
     soc = get_soc_type()
+
     if soc is None:
         return False
 
+    # Check if the SoC is actually a supported RK device
+    # This prevents false positives on non-RK devices (e.g., macOS Docker)
+    # where /proc/device-tree/compatible might exist but contain non-RK content
+    if soc not in SUPPORTED_RK_SOCS:
+        logger.debug(
+            f"SoC '{soc}' is not a supported RK device for RKNN conversion. "
+            f"Supported SoCs: {SUPPORTED_RK_SOCS}"
+        )
+        return False
+
     if not model_type:
         model_type = get_rknn_model_type(model_path)
 

web/public/locales/en/views/classificationModel.json

@@ -1,7 +1,9 @@
 {
   "documentTitle": "Classification Models - Frigate",
   "details": {
-    "scoreInfo": "Score represents the average classification confidence across all detections of this object."
+    "scoreInfo": "Score represents the average classification confidence across all detections of this object.",
+    "none": "None",
+    "unknown": "Unknown"
   },
   "button": {
     "deleteClassificationAttempts": "Delete Classification Images",
@@ -72,7 +74,7 @@
   },
   "renameCategory": {
     "title": "Rename Class",
-    "desc": "Enter a new name for {{name}}. You will be required to retrain the model for the name change to take affect."
+    "desc": "Enter a new name for {{name}}. You will be required to retrain the model for the name change to take effect."
   },
   "description": {
     "invalidName": "Invalid name. Names can only include letters, numbers, spaces, apostrophes, underscores, and hyphens."
@@ -83,7 +85,6 @@
     "aria": "Select Recent Classifications"
   },
   "categories": "Classes",
-  "none": "None",
   "createCategory": {
     "new": "Create New Class"
   },
@@ -138,6 +139,7 @@
         "nameOnlyNumbers": "Model name cannot contain only numbers",
         "classRequired": "At least 1 class is required",
         "classesUnique": "Class names must be unique",
+        "noneNotAllowed": "The class 'none' is not allowed",
         "stateRequiresTwoClasses": "State models require at least 2 classes",
         "objectLabelRequired": "Please select an object label",
         "objectTypeRequired": "Please select a classification type"

web/public/locales/en/views/settings.json

@@ -679,7 +679,7 @@
       "desc": "Manage this Frigate instance's user accounts."
     },
     "addUser": "Add User",
-    "updatePassword": "Update Password",
+    "updatePassword": "Reset Password",
     "toast": {
       "success": {
         "createUser": "User {{user}} created successfully",
@@ -700,7 +700,7 @@
       "role": "Role",
       "noUsers": "No users found.",
       "changeRole": "Change user role",
-      "password": "Password",
+      "password": "Reset Password",
       "deleteUser": "Delete user"
     },
     "dialog": {

web/src/App.tsx

@@ -14,6 +14,7 @@ import ProtectedRoute from "@/components/auth/ProtectedRoute";
 import { AuthProvider } from "@/context/auth-context";
 import useSWR from "swr";
 import { FrigateConfig } from "./types/frigateConfig";
+import ActivityIndicator from "@/components/indicators/activity-indicator";
 
 const Live = lazy(() => import("@/pages/Live"));
 const Events = lazy(() => import("@/pages/Events"));
@@ -50,6 +51,13 @@ function DefaultAppView() {
   const { data: config } = useSWR<FrigateConfig>("config", {
     revalidateOnFocus: false,
   });
+
+  // Compute required roles for main routes, ensuring we have config first
+  // to prevent race condition where custom roles are temporarily unavailable
+  const mainRouteRoles = config?.auth?.roles
+    ? Object.keys(config.auth.roles)
+    : undefined;
+
   return (
     <div className="size-full overflow-hidden">
       {isDesktop && <Sidebar />}
@@ -68,13 +76,11 @@ function DefaultAppView() {
           <Routes>
             <Route
               element={
-                <ProtectedRoute
-                  requiredRoles={
-                    config?.auth.roles
-                      ? Object.keys(config.auth.roles)
-                      : ["admin", "viewer"]
-                  }
-                />
+                mainRouteRoles ? (
+                  <ProtectedRoute requiredRoles={mainRouteRoles} />
+                ) : (
+                  <ActivityIndicator className="absolute left-1/2 top-1/2 -translate-x-1/2 -translate-y-1/2" />
+                )
               }
             >
               <Route index element={<Live />} />

web/src/components/Statusbar.tsx

@@ -116,10 +116,10 @@ export default function Statusbar() {
           }
 
           return (
-            <Link key={gpuTitle} to="/system#general">
+            <Link key={name} to="/system#general">
               {" "}
               <div
-                key={gpuTitle}
+                key={name}
                 className="flex cursor-pointer items-center gap-2 text-sm hover:underline"
               >
                 <MdCircle

web/src/components/card/ClassificationCard.tsx

@@ -4,8 +4,8 @@ import { cn } from "@/lib/utils";
 import {
   ClassificationItemData,
   ClassificationThreshold,
+  ClassifiedEvent,
 } from "@/types/classification";
-import { Event } from "@/types/event";
 import { forwardRef, useMemo, useRef, useState } from "react";
 import { isDesktop, isIOS, isMobile, isMobileOnly } from "react-device-detect";
 import { useTranslation } from "react-i18next";
@@ -40,6 +40,7 @@ type ClassificationCardProps = {
   data: ClassificationItemData;
   threshold?: ClassificationThreshold;
   selected: boolean;
+  clickable: boolean;
   i18nLibrary: string;
   showArea?: boolean;
   count?: number;
@@ -56,6 +57,7 @@ export const ClassificationCard = forwardRef<
     data,
     threshold,
     selected,
+    clickable,
     i18nLibrary,
     showArea = true,
     count,
@@ -101,11 +103,12 @@ export const ClassificationCard = forwardRef<
     <div
       ref={ref}
       className={cn(
-        "relative flex size-full cursor-pointer flex-col overflow-hidden rounded-lg outline outline-[3px]",
+        "relative flex size-full flex-col overflow-hidden rounded-lg outline outline-[3px]",
         className,
         selected
           ? "shadow-selected outline-selected"
           : "outline-transparent duration-500",
+        clickable && "cursor-pointer",
       )}
       onClick={(e) => {
         const isMeta = e.metaKey || e.ctrlKey;
@@ -160,8 +163,12 @@ export const ClassificationCard = forwardRef<
             data.score != undefined ? "text-xs" : "text-sm",
           )}
         >
-          <div className="smart-capitalize">
-            {data.name == "unknown" ? t("details.unknown") : data.name}
+          <div className="break-all smart-capitalize">
+            {data.name == "unknown"
+              ? t("details.unknown")
+              : data.name == "none"
+                ? t("details.none")
+                : data.name}
           </div>
           {data.score != undefined && (
             <div
@@ -186,7 +193,7 @@ export const ClassificationCard = forwardRef<
 
 type GroupedClassificationCardProps = {
   group: ClassificationItemData[];
-  event?: Event;
+  classifiedEvent?: ClassifiedEvent;
   threshold?: ClassificationThreshold;
   selectedItems: string[];
   i18nLibrary: string;
@@ -197,7 +204,7 @@ type GroupedClassificationCardProps = {
 };
 export function GroupedClassificationCard({
   group,
-  event,
+  classifiedEvent,
   threshold,
   selectedItems,
   i18nLibrary,
@@ -232,14 +239,15 @@ export function GroupedClassificationCard({
     const bestTyped: ClassificationItemData = best;
     return {
       ...bestTyped,
-      name: event
-        ? event.sub_label && event.sub_label !== "none"
-          ? event.sub_label
-          : t(noClassificationLabel)
-        : bestTyped.name,
-      score: event?.data?.sub_label_score,
+      name:
+        classifiedEvent?.label && classifiedEvent.label !== "none"
+          ? classifiedEvent.label
+          : classifiedEvent
+            ? t(noClassificationLabel)
+            : bestTyped.name,
+      score: classifiedEvent?.score,
     };
-  }, [group, event, noClassificationLabel, t]);
+  }, [group, classifiedEvent, noClassificationLabel, t]);
 
   const bestScoreStatus = useMemo(() => {
     if (!bestItem?.score || !threshold) {
@@ -284,6 +292,7 @@ export function GroupedClassificationCard({
         data={bestItem}
         threshold={threshold}
         selected={selectedItems.includes(bestItem.filename)}
+        clickable={true}
         i18nLibrary={i18nLibrary}
         count={group.length}
         onClick={(_, meta) => {
@@ -325,36 +334,38 @@ export function GroupedClassificationCard({
                 )}
               >
                 <ContentTitle className="flex items-center gap-2 font-normal capitalize">
-                  {event?.sub_label && event.sub_label !== "none"
-                    ? event.sub_label
+                  {classifiedEvent?.label && classifiedEvent.label !== "none"
+                    ? classifiedEvent.label
                     : t(noClassificationLabel)}
-                  {event?.sub_label && event.sub_label !== "none" && (
-                    <div className="flex items-center gap-1">
-                      <div
-                        className={cn(
-                          "",
-                          bestScoreStatus == "match" && "text-success",
-                          bestScoreStatus == "potential" && "text-orange-400",
-                          bestScoreStatus == "unknown" && "text-danger",
-                        )}
-                      >{`${Math.round((event.data.sub_label_score || 0) * 100)}%`}</div>
-                      <Popover>
-                        <PopoverTrigger asChild>
-                          <button
-                            className="focus:outline-none"
-                            aria-label={t("details.scoreInfo", {
-                              ns: i18nLibrary,
-                            })}
-                          >
-                            <LuInfo className="size-3" />
-                          </button>
-                        </PopoverTrigger>
-                        <PopoverContent className="w-80 text-sm">
-                          {t("details.scoreInfo", { ns: i18nLibrary })}
-                        </PopoverContent>
-                      </Popover>
-                    </div>
-                  )}
+                  {classifiedEvent?.label &&
+                    classifiedEvent.label !== "none" &&
+                    classifiedEvent.score !== undefined && (
+                      <div className="flex items-center gap-1">
+                        <div
+                          className={cn(
+                            "",
+                            bestScoreStatus == "match" && "text-success",
+                            bestScoreStatus == "potential" && "text-orange-400",
+                            bestScoreStatus == "unknown" && "text-danger",
+                          )}
+                        >{`${Math.round((classifiedEvent.score || 0) * 100)}%`}</div>
+                        <Popover>
+                          <PopoverTrigger asChild>
+                            <button
+                              className="focus:outline-none"
+                              aria-label={t("details.scoreInfo", {
+                                ns: i18nLibrary,
+                              })}
+                            >
+                              <LuInfo className="size-3" />
+                            </button>
+                          </PopoverTrigger>
+                          <PopoverContent className="w-80 text-sm">
+                            {t("details.scoreInfo", { ns: i18nLibrary })}
+                          </PopoverContent>
+                        </Popover>
+                      </div>
+                    )}
                 </ContentTitle>
                 <ContentDescription className={cn("", isMobile && "px-2")}>
                   {time && (
@@ -368,14 +379,14 @@ export function GroupedClassificationCard({
               </div>
               {isDesktop && (
                 <div className="flex flex-row justify-between">
-                  {event && (
+                  {classifiedEvent && (
                     <Tooltip>
                       <TooltipTrigger asChild>
                         <div
                           className="cursor-pointer"
                           tabIndex={-1}
                           onClick={() => {
-                            navigate(`/explore?event_id=${event.id}`);
+                            navigate(`/explore?event_id=${classifiedEvent.id}`);
                           }}
                         >
                           <LuSearch className="size-4 text-secondary-foreground" />
@@ -406,6 +417,7 @@ export function GroupedClassificationCard({
                     data={data}
                     threshold={threshold}
                     selected={false}
+                    clickable={false}
                     i18nLibrary={i18nLibrary}
                     onClick={() => {}}
                   >

web/src/components/classification/wizard/Step1NameAndDefine.tsx

@@ -94,7 +94,14 @@ export default function Step1NameAndDefine({
       objectLabel: z.string().optional(),
       objectType: z.enum(["sub_label", "attribute"]).optional(),
       classes: z
-        .array(z.string())
+        .array(
+          z
+            .string()
+            .refine(
+              (val) => val.trim().toLowerCase() !== "none",
+              t("wizard.step1.errors.noneNotAllowed"),
+            ),
+        )
         .min(1, t("wizard.step1.errors.classRequired"))
         .refine(
           (classes) => {
@@ -315,7 +322,7 @@ export default function Step1NameAndDefine({
                       <FormLabel className="text-primary-variant">
                         {t("wizard.step1.classificationType")}
                       </FormLabel>
-                      <Popover>
+                      <Popover modal={true}>
                         <PopoverTrigger asChild>
                           <Button
                             variant="ghost"
@@ -398,7 +405,7 @@ export default function Step1NameAndDefine({
                     ? t("wizard.step1.states")
                     : t("wizard.step1.classes")}
                 </FormLabel>
-                <Popover>
+                <Popover modal={true}>
                   <PopoverTrigger asChild>
                     <Button variant="ghost" size="sm" className="h-4 w-4 p-0">
                       <LuInfo className="size-3" />
@@ -467,6 +474,7 @@ export default function Step1NameAndDefine({
                           )}
                         </div>
                       </FormControl>
+                      <FormMessage />
                     </FormItem>
                   )}
                 />

web/src/components/classification/wizard/Step3ChooseExamples.tsx

@@ -141,7 +141,37 @@ export default function Step3ChooseExamples({
       );
       await Promise.all(categorizePromises);
 
-      // Step 2.5: Create empty folders for classes that don't have any images
+      // Step 2.5: Delete any unselected images from train folder
+      // For state models, all images must be classified, so unselected images should be removed
+      // For object models, unselected images are assigned to "none" so they're already categorized
+      if (step1Data.modelType === "state") {
+        try {
+          // Fetch current train images to see what's left after categorization
+          const trainImagesResponse = await axios.get<string[]>(
+            `/classification/${step1Data.modelName}/train`,
+          );
+          const remainingTrainImages = trainImagesResponse.data || [];
+
+          const categorizedImageNames = new Set(Object.keys(classifications));
+          const unselectedImages = remainingTrainImages.filter(
+            (imageName) => !categorizedImageNames.has(imageName),
+          );
+
+          if (unselectedImages.length > 0) {
+            await axios.post(
+              `/classification/${step1Data.modelName}/train/delete`,
+              {
+                ids: unselectedImages,
+              },
+            );
+          }
+        } catch (error) {
+          // Silently fail - unselected images will remain but won't cause issues
+          // since the frontend filters out images that don't match expected format
+        }
+      }
+
+      // Step 2.6: Create empty folders for classes that don't have any images
       // This ensures all classes are available in the dataset view later
       const classesWithImages = new Set(
         Object.values(classifications).filter((c) => c && c !== "none"),
@@ -156,15 +186,17 @@ export default function Step3ChooseExamples({
       await Promise.all(emptyFolderPromises);
 
       // Step 3: Determine if we should train
-      // For state models, we need ALL states to have examples
-      // For object models, we need at least 2 classes with images
+      // For state models, we need ALL states to have examples (at least 2 states)
+      // For object models, we need at least 1 class with images (the rest go to "none")
       const allStatesHaveExamplesForTraining =
         step1Data.modelType !== "state" ||
         step1Data.classes.every((className) =>
           classesWithImages.has(className),
         );
       const shouldTrain =
-        allStatesHaveExamplesForTraining && classesWithImages.size >= 2;
+        step1Data.modelType === "object"
+          ? classesWithImages.size >= 1
+          : allStatesHaveExamplesForTraining && classesWithImages.size >= 2;
 
       // Step 4: Kick off training only if we have enough classes with images
       if (shouldTrain) {

web/src/components/overlay/ClassificationSelectionDialog.tsx

@@ -132,7 +132,7 @@ export default function ClassificationSelectionDialog({
                   onClick={() => onCategorizeImage(category)}
                 >
                   {category === "none"
-                    ? t("none")
+                    ? t("details.none")
                     : category.replaceAll("_", " ")}
                 </SelectorItem>
               ))}

web/src/components/overlay/ExportDialog.tsx

@@ -440,6 +440,7 @@ function CustomTimeSelector({
       <FaCalendarAlt />
       <div className="flex flex-wrap items-center">
         <Popover
+          modal={false}
           open={startOpen}
           onOpenChange={(open) => {
             if (!open) {
@@ -461,7 +462,10 @@ function CustomTimeSelector({
               {formattedStart}
             </Button>
           </PopoverTrigger>
-          <PopoverContent className="flex flex-col items-center">
+          <PopoverContent
+            disablePortal={isDesktop}
+            className="flex flex-col items-center"
+          >
             <TimezoneAwareCalendar
               timezone={config?.ui.timezone}
               selectedDay={new Date(startTime * 1000)}
@@ -506,6 +510,7 @@ function CustomTimeSelector({
         </Popover>
         <FaArrowRight className="size-4 text-primary" />
         <Popover
+          modal={false}
           open={endOpen}
           onOpenChange={(open) => {
             if (!open) {
@@ -527,7 +532,10 @@ function CustomTimeSelector({
               {formattedEnd}
             </Button>
           </PopoverTrigger>
-          <PopoverContent className="flex flex-col items-center">
+          <PopoverContent
+            disablePortal={isDesktop}
+            className="flex flex-col items-center"
+          >
             <TimezoneAwareCalendar
               timezone={config?.ui.timezone}
               selectedDay={new Date(endTime * 1000)}
@@ -545,7 +553,7 @@ function CustomTimeSelector({
             <SelectSeparator className="bg-secondary" />
             <input
               className="text-md mx-4 w-full border border-input bg-background p-1 text-secondary-foreground hover:bg-accent hover:text-accent-foreground dark:[color-scheme:dark]"
-              id="startTime"
+              id="endTime"
               type="time"
               value={endClock}
               step={isIOS ? "60" : "1"}

web/src/components/overlay/detail/DetailActionsMenu.tsx

@@ -49,6 +49,29 @@ export default function DetailActionsMenu({
     search.data?.type === "audio" ? null : [`review/event/${search.id}`],
   );
 
+  // don't render menu at all if no options are available
+  const hasSemanticSearchOption =
+    config?.semantic_search.enabled &&
+    setSimilarity !== undefined &&
+    search.data?.type === "object";
+
+  const hasReviewItem = !!(reviewItem && reviewItem.id);
+
+  const hasAdminTriggerOption =
+    isAdmin &&
+    config?.semantic_search.enabled &&
+    search.data?.type === "object";
+
+  if (
+    !search.has_snapshot &&
+    !search.has_clip &&
+    !hasSemanticSearchOption &&
+    !hasReviewItem &&
+    !hasAdminTriggerOption
+  ) {
+    return null;
+  }
+
   return (
     <DropdownMenu open={isOpen} onOpenChange={setIsOpen}>
       <DropdownMenuTrigger>

web/src/components/overlay/dialog/TrainFilterDialog.tsx

@@ -170,7 +170,9 @@ export function ClassFilterContent({
                 <FilterSwitch
                   key={item}
                   label={
-                    item === "none" ? t("none") : item.replaceAll("_", " ")
+                    item === "none"
+                      ? t("details.none", { ns: "views/classificationModel" })
+                      : item.replaceAll("_", " ")
                   }
                   isChecked={classes?.includes(item) ?? false}
                   onCheckedChange={(isChecked) => {

web/src/components/settings/ObjectMaskEditPane.tsx

@@ -178,6 +178,19 @@ export default function ObjectMaskEditPane({
         filteredMask.splice(index, 0, coordinates);
       }
 
+      // prevent duplicating global masks under specific object filters
+      if (!globalMask) {
+        const globalObjectMasksArray = Array.isArray(cameraConfig.objects.mask)
+          ? cameraConfig.objects.mask
+          : cameraConfig.objects.mask
+            ? [cameraConfig.objects.mask]
+            : [];
+
+        filteredMask = filteredMask.filter(
+          (mask) => !globalObjectMasksArray.includes(mask),
+        );
+      }
+
       queryString = filteredMask
         .map((pointsArray) => {
           const coordinates = flattenPoints(parseCoordinates(pointsArray)).join(

web/src/components/timeline/DetailStream.tsx

@@ -345,9 +345,9 @@ function ReviewGroup({
   }
 
   const reviewInfo = useMemo(() => {
-    const objectCount = fetchedEvents
-      ? fetchedEvents.length
-      : (review.data.objects ?? []).length;
+    const detectionsCount =
+      review.data?.detections?.length ?? (review.data?.objects ?? []).length;
+    const objectCount = fetchedEvents ? fetchedEvents.length : detectionsCount;
 
     return `${t("detail.trackedObject", { count: objectCount })}`;
   }, [review, t, fetchedEvents]);

web/src/hooks/use-camera-live-mode.ts

@@ -54,7 +54,7 @@ export default function useCameraLiveMode(
   }>({});
 
   useEffect(() => {
-    if (!cameras) return;
+    if (!cameras || cameras.length === 0) return;
 
     const mseSupported =
       "MediaSource" in window || "ManagedMediaSource" in window;

web/src/pages/FaceLibrary.tsx

@@ -68,7 +68,10 @@ import {
   ClassificationCard,
   GroupedClassificationCard,
 } from "@/components/card/ClassificationCard";
-import { ClassificationItemData } from "@/types/classification";
+import {
+  ClassificationItemData,
+  ClassifiedEvent,
+} from "@/types/classification";
 
 export default function FaceLibrary() {
   const { t } = useTranslation(["views/faceLibrary"]);
@@ -922,10 +925,22 @@ function FaceAttemptGroup({
     [onRefresh, t],
   );
 
+  // Create ClassifiedEvent from Event (face recognition uses sub_label)
+  const classifiedEvent: ClassifiedEvent | undefined = useMemo(() => {
+    if (!event || !event.sub_label || event.sub_label === "none") {
+      return undefined;
+    }
+    return {
+      id: event.id,
+      label: event.sub_label,
+      score: event.data?.sub_label_score,
+    };
+  }, [event]);
+
   return (
     <GroupedClassificationCard
       group={group}
-      event={event}
+      classifiedEvent={classifiedEvent}
       threshold={threshold}
       selectedItems={selectedFaces}
       i18nLibrary="views/faceLibrary"
@@ -1011,6 +1026,7 @@ function FaceGrid({
               filepath: `clips/faces/${pageToggle}/${image}`,
             }}
             selected={selectedFaces.includes(image)}
+            clickable={selectedFaces.length > 0}
             i18nLibrary="views/faceLibrary"
             onClick={(data, meta) => onClickFaces([data.filename], meta)}
           >

web/src/pages/Settings.tsx

@@ -437,7 +437,7 @@ export default function Settings() {
 
   return (
     <div className="flex h-full flex-col">
-      <div className="flex items-center justify-between border-b border-secondary p-3">
+      <div className="flex min-h-16 items-center justify-between border-b border-secondary p-3">
         <Heading as="h3" className="mb-0">
           {t("menu.settings", { ns: "common" })}
         </Heading>

web/src/types/classification.ts

@@ -21,6 +21,12 @@ export type ClassificationThreshold = {
   unknown: number;
 };
 
+export type ClassifiedEvent = {
+  id: string;
+  label?: string;
+  score?: number;
+};
+
 export type ClassificationDatasetResponse = {
   categories: {
     [id: string]: string[];

web/src/types/event.ts

@@ -24,5 +24,12 @@ export interface Event {
     type: "object" | "audio" | "manual";
     recognized_license_plate?: string;
     path_data: [number[], number][];
+    // Allow arbitrary keys for attributes (e.g., model_name, model_name_score)
+    [key: string]:
+      | number
+      | number[]
+      | string
+      | [number[], number][]
+      | undefined;
   };
 }

web/src/views/classification/ModelTrainingView.tsx

@@ -62,6 +62,7 @@ import useApiFilter from "@/hooks/use-api-filter";
 import {
   ClassificationDatasetResponse,
   ClassificationItemData,
+  ClassifiedEvent,
   TrainFilter,
 } from "@/types/classification";
 import {
@@ -707,7 +708,7 @@ function LibrarySelector({
                 className="flex-grow cursor-pointer capitalize"
                 onClick={() => setPageToggle(id)}
               >
-                {id === "none" ? t("none") : id.replaceAll("_", " ")}
+                {id === "none" ? t("details.none") : id.replaceAll("_", " ")}
                 <span className="ml-2 text-muted-foreground">
                   ({dataset?.[id].length})
                 </span>
@@ -803,6 +804,7 @@ function DatasetGrid({
               name: "",
             }}
             showArea={false}
+            clickable={selectedImages.length > 0}
             selected={selectedImages.includes(image)}
             i18nLibrary="views/classificationModel"
             onClick={(data, _) => onClickImages([data.filename], true)}
@@ -866,6 +868,12 @@ function TrainGrid({
           };
         })
         .filter((data) => {
+          // Ignore images that don't match the expected format (event-camera-timestamp-state-score.webp)
+          // Expected format has 5 parts when split by "-", and score should be a valid number
+          if (data.score === undefined || isNaN(data.score) || !data.name) {
+            return false;
+          }
+
           if (!trainFilter) {
             return true;
           }
@@ -955,6 +963,7 @@ function StateTrainGrid({
             data={data}
             threshold={threshold}
             selected={selectedImages.includes(data.filename)}
+            clickable={selectedImages.length > 0}
             i18nLibrary="views/classificationModel"
             showArea={false}
             onClick={(data, meta) => onClickImages([data.filename], meta)}
@@ -1027,6 +1036,45 @@ function ObjectTrainGrid({
     };
   }, [model]);
 
+  // Helper function to create ClassifiedEvent from Event
+  const createClassifiedEvent = useCallback(
+    (event: Event | undefined): ClassifiedEvent | undefined => {
+      if (!event || !model.object_config) {
+        return undefined;
+      }
+
+      const classificationType = model.object_config.classification_type;
+
+      if (classificationType === "attribute") {
+        // For attribute type, look at event.data[model.name]
+        const attributeValue = event.data[model.name] as string | undefined;
+        const attributeScore = event.data[`${model.name}_score`] as
+          | number
+          | undefined;
+
+        if (attributeValue && attributeValue !== "none") {
+          return {
+            id: event.id,
+            label: attributeValue,
+            score: attributeScore,
+          };
+        }
+      } else {
+        // For sub_label type, use event.sub_label
+        if (event.sub_label && event.sub_label !== "none") {
+          return {
+            id: event.id,
+            label: event.sub_label,
+            score: event.data?.sub_label_score,
+          };
+        }
+      }
+
+      return undefined;
+    },
+    [model],
+  );
+
   // selection
 
   const [selectedEvent, setSelectedEvent] = useState<Event>();
@@ -1089,11 +1137,13 @@ function ObjectTrainGrid({
       >
         {Object.entries(groups).map(([key, group]) => {
           const event = events?.find((ev) => ev.id == key);
+          const classifiedEvent = createClassifiedEvent(event);
+
           return (
             <div key={key} className="aspect-square w-full">
               <GroupedClassificationCard
                 group={group}
-                event={event}
+                classifiedEvent={classifiedEvent}
                 threshold={threshold}
                 selectedItems={selectedImages}
                 i18nLibrary="views/classificationModel"

web/src/views/live/LiveCameraView.tsx

@@ -147,10 +147,11 @@ export default function LiveCameraView({
 
   // supported features
 
-  const [streamName, setStreamName] = useUserPersistence<string>(
-    `${camera.name}-stream`,
-    Object.values(camera.live.streams)[0],
-  );
+  const [streamName, setStreamName, streamNameLoaded] =
+    useUserPersistence<string>(
+      `${camera.name}-stream`,
+      Object.values(camera.live.streams)[0],
+    );
 
   const isRestreamed = useMemo(
     () =>
@@ -159,6 +160,19 @@ export default function LiveCameraView({
     [config, streamName],
   );
 
+  // validate stored stream name and reset if now invalid
+
+  useEffect(() => {
+    if (!streamNameLoaded) return;
+
+    const available = Object.values(camera.live.streams || {});
+    if (available.length === 0) return;
+
+    if (streamName != null && !available.includes(streamName)) {
+      setStreamName(available[0]);
+    }
+  }, [streamNameLoaded, camera.live.streams, streamName, setStreamName]);
+
   const { data: cameraMetadata } = useSWR<LiveStreamMetadata>(
     isRestreamed ? `go2rtc/streams/${streamName}` : null,
     {