frigate: [Support]: Debug "stuck" detection events

Describe the problem you are having

Hi, I’m using Home Assistant to send notifications on Frigate MQTT events. The issue is that sometimes these MQTT events seem to get “stuck” on a frame and repeat for many hours, even though the object has long since left the camera view.

{ "before": { "id": "1655815890.497768-jqe4g3", "camera": "side_door", "frame_time": 1655830427.443526, "snapshot_time": 1655815992.314924, "label": "person", "top_score": 0.84375, "false_positive": false, "start_time": 1655815890.497768, "end_time": null, "score": 0.76953125, "box": [ 396, 91, 454, 203 ], "area": 6496, "region": [ 220, 0, 588, 368 ], "stationary": true, "motionless_count": 72113, "position_changes": 2, "current_zones": [], "entered_zones": [], "has_clip": true, "has_snapshot": true }, "after": { "id": "1655815890.497768-jqe4g3", "camera": "side_door", "frame_time": 1655830487.946631, "snapshot_time": 1655815992.314924, "label": "person", "top_score": 0.84375, "false_positive": false, "start_time": 1655815890.497768, "end_time": null, "score": 0.76953125, "box": [ 396, 91, 454, 203 ], "area": 6496, "region": [ 220, 0, 588, 368 ], "stationary": true, "motionless_count": 72415, "position_changes": 2, "current_zones": [], "entered_zones": [], "has_clip": true, "has_snapshot": true }, "type": "update" }

Note that the “snapshot_time” is several hours in the past, and so Frigate thinks the object is still present and stationary. What can cause this to happen?

Version

0.10.1 – but this has happened with prior versions as well

Frigate config file

cameras:
  ffmpeg:
      inputs:
        - path: ...
          roles:
           - record
        - path: ...
          roles:
           - detect
      output_args:
        detect: -vf transpose=1 -f rawvideo -pix_fmt yuv420p
    detect:
      width: 720
      height: 1280

Relevant log output

Nothing relevant

FFprobe output from your camera

Nothing relevant

Frigate stats

{"detection_fps":0.0,"detectors":{"cpu":{"detection_start":0.0,"inference_speed":114.84,"pid":225}},"front_door":{"camera_fps":5.1,"capture_pid":232,"detection_fps":0.0,"pid":228,"process_fps":5.1,"skipped_fps":0.0},"mudroom":{"camera_fps":5.1,"capture_pid":233,"detection_fps":0.0,"pid":230,"process_fps":5.1,"skipped_fps":0.0},"service":{"storage":{"/dev/shm":{"free":61.5,"mount_type":"tmpfs","total":67.1,"used":5.6},"/media/frigate/clips":{"free":1592313.5,"mount_type":"ext4","total":1964006.5,"used":371571.3},"/media/frigate/recordings":{"free":1592313.5,"mount_type":"ext4","total":1964006.5,"used":371571.3},"/tmp/cache":{"free":339914.4,"mount_type":"btrfs","total":456159.6,"used":116245.1}},"temperatures":{},"uptime":783516,"version":"0.10.1-83481af"},"side_door":{"camera_fps":5.1,"capture_pid":234,"detection_fps":0.0,"pid":231,"process_fps":5.1,"skipped_fps":0.0}}

Operating system

Other Linux

Install method

Docker CLI

Coral version

CPU (no coral)

Network connection

Wired

Camera make and model

Dahua N43AL52

Any other information that may be helpful

No response