Waste_Bin detection #17360

JDogg016 · 2025-03-25T13:15:08Z

JDogg016
Mar 25, 2025

This is a newbie question, so I apologize if it has been addressed differently than I have seen (or likely did not understand). I have Frigate+ running on an edge Coral computer, and it is really cool. I subscribed to Frigate+ because of its advanced object detection capability.

One particular detection I want to learn more about is waste_bins. I set my waste_bin in a specific area when taking out the trash. However, Frigate+ does not currently identify the waste_bin. My question is, how do I train Frigate+ to apply the waste_bin label? I haven't seen anything like a simple "this is how you train a certain model."

I see where I can audit an image to select areas within it and apply certain labels (i.e., dog, cat, package), but I do not see waste_bin. Am I missing something? Any help would be appreciated.

Here is my config file. For reference if you scroll to the 'Gate Camera' you will see the detection for waste_bin:

mqtt:
  # Optional: Enable mqtt server (default: shown below)
  enabled: true
  # Required: host name
  host: 192.168.7.235
  # Optional: port (default: shown below)
  port: 1883
  # Optional: topic prefix (default: shown below)
  # NOTE: must be unique if you are running multiple instances
  topic_prefix: frigate
  # Optional: client id (default: shown below)
  # NOTE: must be unique if you are running multiple instances
  client_id: frigate
  # Optional: user
  # NOTE: MQTT user can be specified with an environment variable or docker secrets that must begin with 'FRIGATE_'.
  #       e.g. user: '{FRIGATE_MQTT_USER}'
  user: JDogg016
  # Optional: password
  # NOTE: MQTT password can be specified with an environment variable or docker secrets that must begin with 'FRIGATE_'.
  #       e.g. password: '{FRIGATE_MQTT_PASSWORD}'
  password: Helen143!

go2rtc:
  streams:
    Kitchen_Camera:
      - rtspx://192.168.7.1:7441/08TC6mcj9Xm3CyWg
      - ffmpeg:testcam#audio=opus
    Balcony_Camera:
      - rtsp://admin:[email protected]/h264Preview_01_main
    Balcony_Camera_sub:
      - rtsp://admin:[email protected]/h264Preview_01_sub
    Pool_South_Camera:
      - rtsp://admin:[email protected]/h264Preview_01_main
    Pool_South_Camera_sub:
      - rtsp://admin:[email protected]/h264Preview_01_sub
    Pool_North_Camera:
      - rtsp://admin:[email protected]/h264Preview_01_main
    Pool_North_Camera_sub:
      - rtsp://admin:[email protected]/h264Preview_01_sub
    Pavilion_Camera:
      - rtsp://admin:[email protected]/h264Preview_01_main
    Pavilion_Camera_sub:
      - rtsp://admin:[email protected]/h264Preview_01_sub
    Backyard_Camera:
      - rtspx://192.168.7.1:7441/x2qkasfXn7TOUdGF
      - ffmpeg:testcam#audio=opus
    Chicken_Coop_Camera:
      - rtspx://192.168.7.1:7441/WZJRYKvEjeBHy6hX
      - ffmpeg:testcam#audio=opus
    Driveway_Camera:
      - rtspx://192.168.7.1:7441/mMXcxFQTdMvc1K0i
      - ffmpeg:testcam#audio=opus
    Entryway_Camera:
      - rtspx://192.168.7.1:7441/YqklZnEQMf74aWhE
      - ffmpeg:testcam#audio=opus
    Family_Room_Camera:
      - rtspx://192.168.7.1:7441/Xpm41ceZvGAIkLun
      - ffmpeg:testcam#audio=opus
    Front_Doorbell:
      - rtspx://192.168.7.1:7441/4ARyB02va2Aen4Ki
      - ffmpeg:testcam#audio=opus
    Gate:
      - rtspx://192.168.7.1:7441/vRFtngispDeNfgXu
      - ffmpeg:testcam#audio=opus
    Gym:
      - rtspx://192.168.7.1:7441/qYiRgSV7lz9daAf8
      - ffmpeg:testcam#audio=opus
    Office:
      - rtspx://192.168.7.1:7441/7wBOA78rzdcxYKPL
      - ffmpeg:testcam#audio=opus
    Olivias_Room:
      - rtspx://192.168.7.1:7441/QLpKr0GQ3NsE4QLh
      - ffmpeg:testcam#audio=opus
    Staircase:
      - rtspx://192.168.7.1:7441/O7BLEAwwCaWJIWu9
      - ffmpeg:testcam#audio=opus

detectors:
  coral:
    type: edgetpu
    device: usb

semantic_search:
  enabled: true
  reindex: true

detect:
  fps: 5
  enabled: true

cameras:
  Kitchen_Camera: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Kitchen_Camera # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - dog
      mask: 0.51,0.049,0.236,0.998,0.002,0.995,0.001,0.002,0.506,0.007
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      Kitchen_Table:
        coordinates: 0.655,0.439,0.835,0.462,0.995,0.826,0.725,0.856
        loitering_time: 0
    motion:
      threshold: 34
      contour_area: 10
      improve_contrast: true
  Balcony_Camera: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Balcony_Camera
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/Balcony_Camera_sub
          input_args: preset-rtsp-restream
          roles:
            - detect
    objects:
      track:
        - person
        - dog
        - car
        - amazon
        - fedex
        - robot_lawnmower
        - ups
      mask: 0.001,0.003,0.001,0.122,0.999,0.108,0.998,0
      filters:  # Move filters here under `objects`, NOT inside `zones`
        car:
          min_score: 0.4  # Default is usually 0.5, lowering helps detect smaller or partially visible cars
          threshold: 0.5  # Adjust this to reduce false negatives
          min_area: 700  # Minimum pixel area required for a car to be recognized
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      Driveway_Gate_Exit_Opening_Zone:
        coordinates: 0.337,0.172,0.407,0.195,0.364,0.391,0.264,0.381,0.181,0.378
        inertia: 1  # Reduced for quicker detections
        loitering_time: 0
        objects:
          - car
          - person
  Pool_South_Camera: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Pool_South_Camera
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/Pool_South_Camera_sub
          input_args: preset-rtsp-restream
          roles:
            - detect
    objects:
      track:
        - person
        - dog
        - racoon
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      Pool_From_South_Camera:
        coordinates: 
          0.004,0.488,0.03,0.476,0.071,0.47,0.078,0.447,0.117,0.413,0.085,0.342,-0.001,0.377
        inertia: 3
        loitering_time: 0
        objects: person
  Pool_North_Camera: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Pool_North_Camera
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/Pool_North_Camera_sub
          input_args: preset-rtsp-restream
          roles:
            - detect
    objects:
      track:
        - person
        - dog
        - racoon
      mask: 0.002,0.003,0.004,0.21,0.995,0.17,0.998,0.006
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      North_Pool_Occupancy:
        coordinates: 
          0.97,0.272,0.913,0.278,0.798,0.319,0.765,0.347,0.676,0.353,0.611,0.362,0.556,0.424,0.599,0.45,0.913,0.424,0.986,0.468
        inertia: 3
        loitering_time: 0
        objects: person
      Pathway_from_North_Pool_Camera:
        coordinates: 
          0.487,0.271,0.47,0.236,0.58,0.213,0.67,0.198,0.646,0.255,0.587,0.289,0.506,0.325,0.473,0.321,0.581,0.254
        inertia: 3
        loitering_time: 0
        objects: person
  Pavilion_Camera_Camera: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Pavilion_Camera
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/Pavilion_Camera_sub
          input_args: preset-rtsp-restream
          roles:
            - detect
    objects:
      track:
        - person
        - dog
      mask: 0.538,0.045,0.323,0.039,0.32,0.093,0.65,0.094,0.653,0.039
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      Pool:
        coordinates: 
          0.084,0.075,0.141,0.029,0.335,0,0.415,0.005,0.341,0.032,0.348,0.105,0.318,0.123,0.25,0.121,0.237,0.077,0.154,0.114
        loitering_time: 0
        objects: person
  Backyard_Camera: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Backyard_Camera # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - dog
        - robot_lawnmower
        - racoon
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      Playground:
        coordinates: 
          0.575,0.493,0.592,0.544,0.727,0.62,0.717,0.524,0.686,0.493,0.692,0.456,0.651,0.434,0.643,0.463
        loitering_time: 0
        objects: person
        inertia: 3
      Trampoline:
        coordinates: 0.78,0.704,0.859,0.573,0.997,0.664,0.991,0.777,0.991,0.869
        loitering_time: 0
        objects: person
      Lawn_Mower:
        coordinates: 
          0.003,0.44,0.004,0.989,0.992,0.992,0.986,0.887,0.767,0.71,0.58,0.542,0.571,0.48,0.456,0.422,0.374,0.475,0.25,0.398
        loitering_time: 0
        objects: robot_lawnmower
      Backyard_Camera_Pathway:
        coordinates: 0.002,0.42,0,0.263,0.182,0.275,0.18,0.355,0.236,0.392
        inertia: 3
        loitering_time: 0
        objects: person
  Chicken_Coop_Camera: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Chicken_Coop_Camera # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - bird
      mask:
        - 0.107,0.972,0.239,0.321,0.245,0.005,0.003,0.007,0.004,0.983
        - 0.606,0.254,0.997,0.681,0.994,0.023,0.658,0.005
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      Door:
        coordinates: 0.281,0.028,0.264,0.338,0.582,0.392,0.616,0.103
        loitering_time: 0
        objects:
          - bird
          - person
      Coop_Door:
        coordinates: 0.731,0.566,0.69,0.714,0.728,0.794,0.787,0.631
        loitering_time: 0
        objects: bird
  Driveway_Camera: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Driveway_Camera # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - dog
        - car
        - sports_ball
        - racoon
      mask: 0.661,0,0.654,0.082,0.999,0.603,0.998,0.005
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      Soccer:
        coordinates: 0.614,0.275,0.391,0.003,0.471,0.002,0.632,0.195
        loitering_time: 0
        objects: person
      Driveway:
        coordinates: 0.399,0.207,0.003,0.939,0.002,0.993,0.96,0.997,0.9,0.92,0.736,0.613
        loitering_time: 0
        objects:
          - car
          - person
          - sports_ball
        inertia: 3
  Entryway_Camera: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Entryway_Camera # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - dog
      mask: 
        0.001,0.005,0.007,0.519,0.192,0.472,0.283,0.387,0.439,0.345,0.597,0.294,0.784,0.319,0.792,0.003
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
  Family_Room_Camera: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Family_Room_Camera # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - dog
      mask: 0.003,0.528,0.073,0.625,0.211,0.225,0.179,0.108,0.249,0.012,0.002,0.003
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      Couch:
        coordinates: 
          0.094,0.659,0.182,0.636,0.232,0.373,0.296,0.331,0.312,0.244,0.293,0.207,0.197,0.289
        inertia: 3
        loitering_time: 0
      Sofa:
        coordinates: 0.176,0.733,0.155,0.899,0.503,0.861,0.52,0.584,0.209,0.739
        loitering_time: 0
  Front_Doorbell: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Front_Doorbell # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - dog
        - package
        - racoon
      mask: 0.004,0.03,0.321,0.03,0.326,0.122,0.704,0.231,0.997,0.217,0.998,0.003,0.002,0
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      Dog_at_Door:
        coordinates: 
          0.348,0.559,0.356,0.632,0.242,0.977,0.299,0.994,0.846,0.74,0.628,0.593,0.621,0.532
        loitering_time: 0
        objects: dog
      Person_at_Door:
        coordinates: 
          0.355,0.41,0.363,0.623,0.304,0.992,0.241,0.997,0.843,0.739,0.608,0.603,0.58,0.412
        loitering_time: 0
        objects: person
      Package:
        coordinates: 
          0.35,0.561,0.357,0.627,0.246,0.995,0.99,0.996,0.996,0.859,0.631,0.591,0.622,0.534
        loitering_time: 0
        objects: package
        inertia: 3
      Racoon:
        coordinates: 
          0.808,0.544,0.788,0.553,0.682,0.601,0.676,0.57,0.62,0.536,0.369,0.576,0.249,0.999,0.99,0.994,0.996,0.859,0.874,0.758,0.889,0.566
        loitering_time: 0
        objects: racoon
    motion:
      mask: 0.003,0.026,0.324,0.028,0.324,0.001,0.001,0.004
  Gate: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Gate # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - dog
        - car
        - bicycle
        - package
        - waste_bin
        - robot_lawnmower
        - amazon
        - fedex
        - ups
        - racoon
      mask: -0.001,0.328,0.997,0.307,0.997,0.002,0.002,0
      filters:
        person:
          mask:
            - -0.001,0.328,0.997,0.307,0.997,0.002,0.002,0
            - 0,0.001,0.502,0.002,0.485,0.395,0.004,0.397
        waste_bin:
          min_score: 0.3  # Default is usually 0.5, lowering helps detect smaller or partially visible cars
          threshold: 0.5  # Adjust this to reduce false negatives
          min_area: 500

    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      Street:
        coordinates: 0.448,0.508,0.53,0.501,0.72,0.508,1,0.527,0.997,0.707,0.658,0.591
        loitering_time: 0
        inertia: 3
      Trash_Cans:
        coordinates: 
          0.436,0.56,0.436,0.541,0.523,0.524,0.551,0.518,0.604,0.52,0.658,0.595,0.444,0.645
        loitering_time: 0
        objects: waste_bin
        inertia: 3
      Gate_Entry_Driveway:
        coordinates: 0,0.429,0.553,0.421,0.998,0.881,0.998,0.995,0.002,0.992
        inertia: 3
        loitering_time: 0
        objects:
          - car
          - dog
          - person
      Front_Gate_Lawn_Mower:
        coordinates: 0.32,0.631,0.646,0.598,0.482,0.523,0.415,0.531
        inertia: 3
        loitering_time: 0
        objects: robot_lawnmower
  Gym: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Gym # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - dog
      mask: 
        0.002,0.005,0.003,0.352,0.587,0.169,0.587,0.127,0.675,0.098,0.7,0.071,0.998,0.174,0.996,0.005
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
  Office: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Office # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - dog
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
    zones:
      Helens_Desk:
        coordinates: 0.863,0.634,0.788,0.571,0.747,0.655,0.819,0.732
        loitering_time: 0
        objects: person
      Justins_Desk:
        coordinates: 0.681,0.991,0.461,0.998,0.443,0.904,0.64,0.909
        loitering_time: 0
        objects: person
  Olivias_Room: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Olivias_Room # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - dog
      mask: 
        0.006,0.275,0.046,0.26,0.052,0.296,0.247,0.27,0.466,0.274,0.593,0.267,0.764,0.282,0.997,0.328,0.995,0.017,0.003,0.009
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5
  Staircase: # <------ Name the camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Staircase # <----- The stream you want to use for detection
          input_args: preset-rtsp-restream

          roles:
            - detect
    objects:
      track:
        - person
        - dog
      mask:
        - 0.046,0.463,0.394,0.288,0.398,0,0.024,0.005,0.002,0.997,0.133,0.99
        - 0.36,0.002,0.354,0.2,0.396,0.22,0.395,0.328,0.494,0.319,0.499,0.202,0.76,0.002
    snapshots:
      enabled: true
      timestamp: true
      bounding_box: true
      retain:
        default: 5

    zones:
      Stairs:
        coordinates: 0.941,0.075,0.396,0.495,0.29,0.758,0.31,0.885,0.333,0.889,0.871,0.746
        loitering_time: 0
version: 0.15-1

Answered by hawkeye217

Mar 25, 2025

Right, you need to go to the Cameras item on the left menu, click Edit, and then check the applicable labels for that camera.

View full answer

hawkeye217 · 2025-03-25T13:27:33Z

hawkeye217
Mar 25, 2025
Collaborator

Have you seen the official documentation on how to improve your model? https://docs.frigate.video/plus/improving_model

0 replies

JDogg016 · 2025-03-25T13:41:50Z

JDogg016
Mar 25, 2025
Author

Yes. My point is that within the ability to annotate (for example) there is no option to identify the waste_bin in this image as a waste_bin:

5 replies

hawkeye217 Mar 25, 2025
Collaborator

Right, you need to go to the Cameras item on the left menu, click Edit, and then check the applicable labels for that camera.

Answer selected by NickM-27

JDogg016 Mar 25, 2025
Author

Thank you so much! I did not realize that ability existed and I apologize if I missed it somewhere else in the documentation. Hopefully that will allow me to better annotate damages going forward.

hawkeye217 Mar 25, 2025
Collaborator

No problem, happy to help.

JDogg016 Mar 25, 2025
Author

One last question, since while on the subject. I want to have the ability to detect a sports ball within my driveway.

Sports_ball is one of the list of available objects. Is there a best practice for annotating these and training sports balls (ie a basketball)?

hawkeye217 Mar 25, 2025
Collaborator

sports_ball is an object that the default free model detects, but it is not an object that Frigate+ models detect. You can't combine models or labels.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

Waste_Bin detection #17360

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{editor}}'s edit

{{editor}}'s edit

Uh oh!

Replies: 2 comments 5 replies

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Select a reply

Uh oh!

Uh oh!

Waste_Bin detection #17360

Uh oh!

Uh oh!

JDogg016 Mar 25, 2025

Replies: 2 comments · 5 replies

Uh oh!

hawkeye217 Mar 25, 2025 Collaborator

Uh oh!

JDogg016 Mar 25, 2025 Author

Uh oh!

hawkeye217 Mar 25, 2025 Collaborator

Uh oh!

JDogg016 Mar 25, 2025 Author

Uh oh!

hawkeye217 Mar 25, 2025 Collaborator

Uh oh!

JDogg016 Mar 25, 2025 Author

Uh oh!

hawkeye217 Mar 25, 2025 Collaborator

JDogg016
Mar 25, 2025

Replies: 2 comments 5 replies

hawkeye217
Mar 25, 2025
Collaborator

JDogg016
Mar 25, 2025
Author

hawkeye217 Mar 25, 2025
Collaborator

JDogg016 Mar 25, 2025
Author

hawkeye217 Mar 25, 2025
Collaborator

JDogg016 Mar 25, 2025
Author

hawkeye217 Mar 25, 2025
Collaborator