diff --git a/docs/docs/troubleshooting/faqs.md b/docs/docs/troubleshooting/faqs.md index ff2379ea72..beaf287b6a 100644 --- a/docs/docs/troubleshooting/faqs.md +++ b/docs/docs/troubleshooting/faqs.md @@ -110,3 +110,19 @@ No. Frigate uses the TCP protocol to connect to your camera's RTSP URL. VLC auto TCP ensures that all data packets arrive in the correct order. This is crucial for video recording, decoding, and stream processing, which is why Frigate enforces a TCP connection. UDP is faster but less reliable, as it does not guarantee packet delivery or order, and VLC does not have the same requirements as Frigate. You can still configure Frigate to use UDP by using ffmpeg input args or the preset `preset-rtsp-udp`. See the [ffmpeg presets](/configuration/ffmpeg_presets) documentation. + +### Why does Frigate keep creating new events for my parked car? + +Stationary tracking is designed to _prevent_ this — a parked car should stay one tracked object and not generate new events. If you're getting repeated events for the same car, it's likely that Frigate is losing the tracked object and re-detecting it as a new one. + +Open one of the events in Explore → **Tracking Details**. If the detection scores are low (< 70% or so), the model isn't confident the parked car is a car. This is common with the free [COCO-trained](https://cocodataset.org/#explore) object detection models on steep/top-down angles, partially occluded cars, foliage, or low-light footage. When detections fall below `min_score` for too many frames the tracker loses the object, and the next confident frame creates a brand new one. + +What helps: + +- **Improve the view** — even a small angle change that gets more of the car visible could lift scores enough to stabilize tracking. +- **Use a more accurate model** — `yolov9` over `mobiledet` or a larger version of `yolov9`, like `yolov9-s` over `yolov9-t` (though this will increase inference time and is still trained on the same COCO dataset). Also, [Frigate+](https://frigate.video/plus) for models trained on security-camera footage and tunable to your scene. Users have reported that Frigate+ is by far the most reliable fix. +- **Don't set `detect -> stationary -> max_frames` for `car`** — it artificially ends tracking and forces re-detection as a new object. See [Stationary Objects](../configuration/stationary_objects.md). +- **Restrict alerts to the areas you care about** with `required_zones` — see [Zones](../configuration/zones.md#restricting-alerts-and-detections-to-specific-zones). Make sure those zones use the default `loitering_time: 0` unless you specifically want the review item to stay open until the car leaves. +- **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.