add mse/rtc live view faq

docs/data/object_detectors_models.json

@@ -44,22 +44,22 @@
   "openvino": {
     "title": "OpenVINO",
     "models": [
-      {
-        "key": "ssd",
-        "label": "SSDLite MobileNet v2",
-        "recommended": true,
-        "download": "An OpenVINO model is provided in the container at `/openvino-model/ssdlite_mobilenet_v2.xml` and is used by this detector type by default. The model comes from Intel's Open Model Zoo [SSDLite MobileNet V2](https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/ssdlite_mobilenet_v2) and is converted to an FP16 precision IR model.",
-        "ui": "Navigate to **Settings > System > Detectors and model** and select **OpenVINO** from the detector type dropdown and click **Add**, then set device to `GPU` (or `NPU`). Then on the same page, in the **Custom Model** tab, configure:\n\n| Field                                    | Value                                      |\n| ---------------------------------------- | ------------------------------------------ |\n| **Object detection model input width**   | `300`                                      |\n| **Object detection model input height**  | `300`                                      |\n| **Model Input Tensor Shape**             | `nhwc`                                     |\n| **Model Input Pixel Color Format**       | `bgr`                                      |\n| **Custom object detector model path**    | `/openvino-model/ssdlite_mobilenet_v2.xml` |\n| **Label map for custom object detector** | `/openvino-model/coco_91cl_bkgr.txt`       |",
-        "yaml": "detectors:\n  ov:\n    type: openvino\n    device: GPU # Or NPU\n\nmodel:\n  width: 300\n  height: 300\n  input_tensor: nhwc\n  input_pixel_format: bgr\n  path: /openvino-model/ssdlite_mobilenet_v2.xml\n  labelmap_path: /openvino-model/coco_91cl_bkgr.txt"
-      },
       {
         "key": "yolov9",
         "label": "YOLOv9",
-        "recommended": false,
+        "recommended": true,
         "download": "YOLOv9 model can be exported as ONNX using the command below. You can copy and paste the whole thing to your terminal and execute, altering `MODEL_SIZE=t` and `IMG_SIZE=320` in the first line to the [model size](https://github.com/WongKinYiu/yolov9#performance) you would like to convert (available model sizes are `t`, `s`, `m`, `c`, and `e`, common image sizes are `320` and `640`).\n\n```sh\ndocker build . --build-arg MODEL_SIZE=t --build-arg IMG_SIZE=320 --output . -f- <<'EOF'\nFROM python:3.11 AS build\nRUN apt-get update && apt-get install --no-install-recommends -y cmake libgl1 && rm -rf /var/lib/apt/lists/*\nCOPY --from=ghcr.io/astral-sh/uv:0.10.4 /uv /bin/\nWORKDIR /yolov9\nADD https://github.com/WongKinYiu/yolov9.git .\nRUN uv pip install --system -r requirements.txt\nRUN uv pip install --system onnx==1.18.0 onnxruntime onnx-simplifier==0.4.* onnxscript\nARG MODEL_SIZE\nARG IMG_SIZE\nADD https://github.com/WongKinYiu/yolov9/releases/download/v0.1/yolov9-${MODEL_SIZE}-converted.pt yolov9-${MODEL_SIZE}.pt\nRUN sed -i \"s/ckpt = torch.load(attempt_download(w), map_location='cpu')/ckpt = torch.load(attempt_download(w), map_location='cpu', weights_only=False)/g\" models/experimental.py\nRUN python3 export.py --weights ./yolov9-${MODEL_SIZE}.pt --imgsz ${IMG_SIZE} --simplify --include onnx\nFROM scratch\nARG MODEL_SIZE\nARG IMG_SIZE\nCOPY --from=build /yolov9/yolov9-${MODEL_SIZE}.onnx /yolov9-${MODEL_SIZE}-${IMG_SIZE}.onnx\nEOF\n```",
         "ui": "Navigate to **Settings > System > Detectors and model** and select **OpenVINO** from the detector type dropdown and click **Add**, then set device to `GPU` (or `NPU`). Then on the same page, in the **Custom Model** tab, configure:\n\n| Field                                    | Value                                                    |\n| ---------------------------------------- | -------------------------------------------------------- |\n| **Object Detection Model Type**          | `yolo-generic`                                           |\n| **Object detection model input width**   | `320` (should match the imgsize set during model export) |\n| **Object detection model input height**  | `320` (should match the imgsize set during model export) |\n| **Model Input Tensor Shape**             | `nchw`                                                   |\n| **Model Input D Type**                   | `float`                                                  |\n| **Custom object detector model path**    | `/config/model_cache/yolo.onnx`                          |\n| **Label map for custom object detector** | `/labelmap/coco-80.txt`                                  |",
         "yaml": "detectors:\n  ov:\n    type: openvino\n    device: GPU # or NPU\n\nmodel:\n  model_type: yolo-generic\n  width: 320 # <--- should match the imgsize set during model export\n  height: 320 # <--- should match the imgsize set during model export\n  input_tensor: nchw\n  input_dtype: float\n  path: /config/model_cache/yolo.onnx\n  labelmap_path: /labelmap/coco-80.txt"
       },
+      {
+        "key": "ssd",
+        "label": "SSDLite MobileNet v2",
+        "recommended": false,
+        "download": "An OpenVINO model is provided in the container at `/openvino-model/ssdlite_mobilenet_v2.xml` and is used by this detector type by default. The model comes from Intel's Open Model Zoo [SSDLite MobileNet V2](https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/ssdlite_mobilenet_v2) and is converted to an FP16 precision IR model.",
+        "ui": "Navigate to **Settings > System > Detectors and model** and select **OpenVINO** from the detector type dropdown and click **Add**, then set device to `GPU` (or `NPU`). Then on the same page, in the **Custom Model** tab, configure:\n\n| Field                                    | Value                                      |\n| ---------------------------------------- | ------------------------------------------ |\n| **Object detection model input width**   | `300`                                      |\n| **Object detection model input height**  | `300`                                      |\n| **Model Input Tensor Shape**             | `nhwc`                                     |\n| **Model Input Pixel Color Format**       | `bgr`                                      |\n| **Custom object detector model path**    | `/openvino-model/ssdlite_mobilenet_v2.xml` |\n| **Label map for custom object detector** | `/openvino-model/coco_91cl_bkgr.txt`       |",
+        "yaml": "detectors:\n  ov:\n    type: openvino\n    device: GPU # Or NPU\n\nmodel:\n  width: 300\n  height: 300\n  input_tensor: nhwc\n  input_pixel_format: bgr\n  path: /openvino-model/ssdlite_mobilenet_v2.xml\n  labelmap_path: /openvino-model/coco_91cl_bkgr.txt"
+      },
       {
         "key": "yolo-legacy",
         "label": "YOLO (v3, v4, v7)",

docs/docs/configuration/live.md

@@ -434,3 +434,7 @@ When your browser runs into problems playing back your camera streams, it will l
    ```
 
    The same applies to your `record` stream: if its aspect ratio differs from your `detect` stream, your recordings will appear in a different shape than the live view. For consistent framing across live view and recordings, use the same aspect ratio for all of a camera's streams (the resolution can still differ).
+
+9. **Why does Frigate prefer MSE over WebRTC for live view?**
+
+   Frigate prefers MSE because it delivers a better out-of-the-box experience than WebRTC on nearly every axis that matters for a security camera system. MSE is an open standard optimized and supported by all modern browsers, works without any extra configuration (WebRTC requires port forwarding and candidate setup, and lacks H.265 support in some browsers), and requires no internet access for NAT traversal. More importantly, MSE runs over TCP, so every frame arrives and is decoded in order, so nothing is ever silently skipped. WebRTC optimizes for latency over UDP by discarding late or incomplete frames, which works against you on cellular or spotty Wi-Fi: you can end up with frozen video, visual corruption, or gaps in the feed without ever knowing you missed something. Frigate's enhanced MSE player has adaptive speed playback and has been tuned for latency and connection robustness that meets or exceeds WebRTC, so you get near-real-time playback with a guarantee that when the video plays, every frame is actually there - which, for an NVR whose whole purpose is letting you see what happened, matters more than shaving fractions of a second off a latency number. That's why Frigate defaults to MSE and reserves WebRTC for cases that require it, like two-way talk.

docs/docs/configuration/object_filters.md

@@ -7,7 +7,7 @@ import ConfigTabs from "@site/src/components/ConfigTabs";
 import TabItem from "@theme/TabItem";
 import NavPath from "@site/src/components/NavPath";
 
-There are several types of object filters that can be used to reduce false positive rates.
+There are several types of object filters that can be used to reduce [false positive](/frigate/glossary#false-positive) rates.
 
 ## Object Scores
 
@@ -148,4 +148,4 @@ cameras:
 
 ### Object Masks
 
-[Object Filter Masks](/configuration/masks) are a last resort but can be useful when false positives are in the relatively same place but can not be filtered due to their size or shape. Object filter masks can be configured in <NavPath path="Settings > Camera configuration > Masks / Zones" />.
+[Object Filter Masks](/configuration/masks#object-filter-masks) are a last resort but can be useful when false positives are in the relatively same place but can not be filtered due to their size or shape. Object filter masks can be configured in <NavPath path="Settings > Camera configuration > Masks / Zones" />.

docs/docs/troubleshooting/faqs.md

@@ -140,3 +140,16 @@ What helps:
 - **Filter impossible locations** with [object filter masks](../configuration/masks.md#object-filter-masks) if cars are being detected on rooftops, treetops, etc.
 
 See [Object Filters](../configuration/object_filters.md) for more on tuning `min_score` and `threshold` — note that raising them too high will make this exact problem worse.
+
+### How do I correct Frigate when it detects something as the wrong object?
+
+Frigate's object detection relies on a machine learning [model](../frigate/glossary.md#model), and the free [COCO-trained](https://cocodataset.org/#explore) models that ship with Frigate can misidentify objects in scenes they weren't trained on. There are two ways to handle this, depending on whether you want to _teach_ the model or just _suppress_ the bad result.
+
+**Train or fine-tune a model with your own images.** The most durable fix is to improve the model itself. The biggest gains usually come from fine-tuning a model on images from your own cameras so it learns your specific scene. Some tools are freely available, and [Frigate+](https://frigate.video/plus) is a paid option that does this - models are trained on security-camera footage and can be fine-tuned on images you submit from your own setup. When Frigate mislabels something, open the tracked object in Explore, select the **Snapshot** tab, and use **Submit to Frigate+** to send the example with the correct label (or mark it as a [false positive](../frigate/glossary.md#false-positive)). Once you've submitted examples and [requested a model](../plus/first_model.md), the retrained model will be more accurate for your cameras. See [Submitting examples to Frigate+](../integrations/plus.md#submit-examples) for the full workflow.
+
+**Suppress the misidentification with filters.** You can use filters to stop a specific false positive from being tracked:
+
+- Tune `min_score` / `threshold`, or add `min_area` / `max_area` / `min_ratio` / `max_ratio` filters — see [Object Filters](../configuration/object_filters.md).
+- If the false positive is always in the same fixed spot (like a statue or mailbox that reads as a person), add an [object filter mask](../configuration/masks.md#object-filter-masks) over that location.
+
+Filters and masks only hide the incorrect result - they don't teach Frigate what the object actually is. For that, fine-tune your own model or use Frigate+.