Trying to avoid recording cars on the road especially on the exports.

[vajonam]

I have a camera configured.

  front: # <------ Name the camera
    ffmpeg:
      inputs:
      - path: rtsp:https://frigate:8554/front   # <--- the name here must match the name of the camera in restream
        input_args: preset-rtsp-restream
        roles:
        - record
      - path: 
          rtsp:https://admin:{FRIGATE_RTSP_PASSWORD}@ip_camera01:554/cam/realmonitor?channel=1&subtype=1       # <--- the name here must match the name of the camera in restream
        roles:
        - detect
    detect:
      enabled: true # <---- disable detection until you have a working camera feed
      width: 864  # <---- update for your camera's resolution
      height: 480 # <---- update for your camera's resolution
      fps: 8.0
      stationary:
        max_frames:
          objects:
            person: 300

    motion:
      mask:
      - 830,17,833,41,576,41,575,14 # timestamp
      delta_alpha: 0.4
      improve_contrast: false
      threshold: 45
      contour_area: 25

    snapshots:
      required_zones:
      - driveway
      - yard
      - gargae
      - road
    objects:
      filters:
        car:
          min_score: 0.8
          threshold: 0.85
          mask: [] # 347,251,517,248,524,175,356,182
        person:
          mask: 575,222,576,192,504,192,502,221
          # min_ratio: 0.2
          max_ratio: 0.90
          min_score: 0.78
          # threshold: 0.88
    record:
      events:
        required_zones:
        - driveway
        - yard
        - gargae
        retain:
          default: 7
          mode: active_objects


    zones:
      driveway:
        coordinates: 215,480,854,480,854,400,694,244,527,133,343,159,128,205
        objects:
        - car
      yard:
        coordinates: 854,480,854,157,714,83,367,95,0,145,0,480
        # 790,480,854,256,801,214,780,0,478,0,392,104,486,117,433,246,225,249,33,480
        objects:
        - person
        - bicycle
      road:
        coordinates: 0,48,0,126,220,84,451,64,608,61,803,64,796,0,487,0,180,0
        objects:
        - car
        - bus
        - truck
        filters:
          car:
            threshold: 0.85
            min_area: 11000
      garage:
        coordinates: 191,445,211,480,854,480,854,380,765,310
        objects:
        - car

and a global record section that looks like

record:
  enabled: true
  events:
    pre_capture: 8
    post_capture: 8
    objects:
    - person
    - car
    - bird
    - cat
    - dog
    retain:
      default: 10
      mode: active_objects
      objects:
        person: 15
        car: 15

I wan to capture a snapshot of the car on the road zone, but don't want to record them. However I want to record things like cars on the driveway, person in yard and car in garage. Is this config correct? Looks like I am still seeing car on road in the VOD api m3u8 when exporting the events. I want clean up the exports to only events in the required zones.

[vajonam]

I am using https://frigate.domain.com/vod/front/start/1695268800/end/1695355200/index.m3u8 this API get a list of events. But I guess this is not recording, its events, so it cannot apply that filter I have above?

[vajonam]

To achieve what I want, I can add a mask to remove the car from an active objects once it reaches its parking position.

[NickM-27]

you could be there is always the chance that the box gets detected differently and creates a new event.

[vajonam]

yup got it. the active_objects mislead me, will an enhancement to differentiate between "moving" objects and "stationary" objects be possible in the future?

[NickM-27]

I think there is some confusion, active objects specifically means not stationary objects. However, it means any active objects, not necessarily referring to the object that is in the required zone

[vajonam]

sounds like I cannot achieve what I need with the settings as they are today. seems that required_zones and active_objects should play nice with each other. thanks.

[vajonam]

Looking at the docs.

https://deploy-preview-6262--frigate-docs.netlify.app/configuration/record/#what-do-the-different-retain-modes-mean

Let's consider a scenario where you drive up and park in your driveway, go inside, then come back out 4 hours later.

With the active_objects it would only keep segments where the object was active. In this case the only segments that would be saved would be the ones where the car was driving up, you going inside, you coming outside, and the car driving away. Essentially reducing the 4 hours to a minute or two of event footage.

This is what I want but however its still recording the cars that drive by on the road, what am I doing wrong in my setup?

[vajonam]

sorry if I dont understand, if this is a defect or is if its as designed then does the docs needs an update?

[NickM-27]

yeah, things are working as expected, the docs just need to be tweaked a bit