[Support]: USPS Car detected but labeled as Fedex Car in UI

[rolf-tx]

Checklist

• I have updated to the latest available Frigate version.
• I have cleared the cache of my browser.
• I have tried a different browser to see if it is related to my browser.
• I have tried reproducing the issue in incognito mode to rule out problems with any third party extensions or plugins I have installed.

Describe the problem you are having

This is on 0.15.0-RC1
USPS Car detected but labeled as Fedex Car in UI.
Uploading to Frigate+ show object Detected as USPS Car as expected.

Steps to reproduce

1. N/A

Version

0.15.0-RC1

In which browser(s) are you experiencing the issue with?

No response

Frigate config file

mqtt:
  enabled: true
  host: ha.example.com
  user: frigate-mqtt
  password: xxxxx
database:
  path: /config/frigate.db


go2rtc:
  streams:
    garage:
      - rtsp://rolfcam:[email protected]:554/Streaming/Channels/101/
    driveway:
      - rtsp://rolfcam:[email protected]:554/Streaming/Channels/101/
    frontentry:
      - rtsp://rolfcam:[email protected]:554/Streaming/Channels/101/
    garage_sub:
      - rtsp://rolfcam:[email protected]:554/Streaming/Channels/102/
    driveway_sub:
      - rtsp://rolfcam:[email protected]:554/Streaming/Channels/102/
    frontentry_sub:
      - rtsp://rolfcam:[email protected]:554/Streaming/Channels/102/


cameras:
  garage:
    enabled: true
    ffmpeg:
      output_args:
        record: preset-record-generic-audio-aac
      inputs:
        - path: rtsp://127.0.0.1:8554/garage
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/garage_sub
          input_args: preset-rtsp-restream
          roles:
            - detect
    detect:
      enabled: true
      fps: 6
      width: 640
      height: 480
      annotation_offset: -2500
    birdseye:
      order: 2
    motion:
      mask: 0.518,0.122,0.517,0.069,0.1,0.071,0.037,0.072,0.037,0.122
    objects:
      filters:
        car:
          mask: 0,0,0,1,1,1,1,0


  driveway:
    enabled: true
    ffmpeg:
      output_args:
        record: preset-record-generic-audio-aac
      inputs:
        - path: rtsp://127.0.0.1:8554/driveway
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/driveway_sub
          input_args: preset-rtsp-restream
          roles:
            - detect
    detect:
      enabled: true
      fps: 6
      width: 640
      height: 360
      annotation_offset: -2200
    birdseye:
      order: 3
    motion:
      mask: 0.043,0.069,0.283,0.067,0.284,0.106,0.041,0.105
    zones:
      zone_driveway_pavement:
        coordinates: 
          0.036,0.999,0.117,0.585,0.192,0.442,0.225,0.296,0.544,0.262,0.602,0.298,0.726,0.293,0.85,0.399,0.842,0.473,0.989,0.783,0.991,1,0.499,1
    review:
      alerts:
        required_zones: zone_driveway_pavement


  frontentry:
    enabled: true
    ffmpeg:
      output_args:
        record: preset-record-generic-audio-aac
      inputs:
        - path: rtsp://127.0.0.1:8554/frontentry
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/frontentry_sub
          input_args: preset-rtsp-restream
          roles:
            - detect
    detect:
      enabled: true
      fps: 6
      width: 480
      height: 640
      annotation_offset: -2600
    motion:
      threshold: 50
      mask:
        - 0.131,0.164,0.295,0.146,0.304,0.084,0.131,0.081
        - 0,0.084,0.067,0.083,0.152,0.083,0.306,0.084,0.542,0.089,0.54,0,0.375,0,0,0
        - 0.073,0.167,0,0.155,0,0.661,0.184,0.537
    zones:
      zone_frontentry:
        coordinates: 0.423,1,0.723,0.823,0.725,0.275,0.301,0.247,0,0.964
    birdseye:
      order: 1
    review:
      alerts:
        required_zones: zone_frontentry


objects:
  track:
    - person
    - face
    - car
    - motorcycle
    - bicycle
    - amazon
    - usps
    - fedex
    - ups
    - dog
    - cat
    - package
    - waste_bin
  filters:
    person:
      min_score: 0.7
      threshold: 0.8
    face:
      min_score: 0.6
      threshold: 0.7
    car:
      min_score: 0.7
      threshold: 0.8
    motorcycle:
      min_score: 0.7
      threshold: 0.8
    bicycle:
      min_score: 0.7
      threshold: 0.8
    amazon:
      min_score: 0.6
      threshold: 0.65
    usps:
      min_score: 0.6
      threshold: 0.65
    fedex:
      min_score: 0.6
      threshold: 0.65
    ups:
      min_score: 0.6
      threshold: 0.65
    dog:
      min_score: 0.6
      threshold: 0.7
    cat:
      min_score: 0.6
      threshold: 0.7
    package:
      min_score: 0.6
      threshold: 0.7
    waste_bin:
      min_score: 0.7
      threshold: 0.8
    license_plate:
      min_score: 0.7
      threshold: 0.8

timestamp_style:
  position: tl

detectors:
  coral:
    type: edgetpu
    device: usb
#    type: openvino
#    device: GPU
#  ov_1:
#    type: openvino
#    device: GPU

model:
  path: plus://e8909389a9daa14f88cb1bd968bb2200 #TPU 320x320
#  path: plus://65fda372873dc95736a771e0b4bd10fb #320x320
#  path: plus://a3d58c394707473b7236607a0ea39d81 #640x640 slow



#model:
#  width: 300
#  height: 300
#  input_tensor: nhwc
#  input_pixel_format: bgr
#  path: /openvino-model/ssdlite_mobilenet_v2.xml
#  labelmap_path: /openvino-model/coco_91cl_bkgr.txt



ffmpeg:
  hwaccel_args: preset-intel-qsv-h264
#  hwaccel_args: preset-intel-vaapi
#  hwaccel_args: '""'

record:
  enabled: true
  retain:
    days: 15
    mode: all
  alerts:
    pre_capture: 10
    post_capture: 10
    retain:
      days: 30
      mode: motion
  detections:
    pre_capture: 10
    post_capture: 10
    retain:
      days: 30
      mode: motion
  preview:
    quality: high

snapshots:
  enabled: true
  retain:
    default: 30

# Include all cameras by default in Birdseye view
birdseye:
  enabled: true
  mode: continuous
  width: 1280
  height: 720
  restream: true

version: 0.15-1

semantic_search:
  enabled: false
  reindex: false
  model_size: small

docker-compose file or Docker CLI command

Homeassistant integration

Relevant Frigate log output

N/A

Relevant go2rtc log output

N/A

Operating system

HassOS

Install method

HassOS Addon

Network connection

Wired

Camera make and model

hikvision

Screenshots of the Frigate UI's System metrics pages

N/A

Any other information that may be helpful

No response

[hawkeye217]

You should look at the Object Lifecycle for the Fedex tracked object.

Click the magnifying glass next to "Fedex". That will take you to Explore. Then, click the thumbnail to show the Tracked Object Details pane and then click the Object Lifecycle tab. What does it show as you scroll through each lifecycle event?

[rolf-tx]

Each of the five lifecycle events show FedEx.
Only the snapshot shows USPS.
What does this imply?

[hawkeye217]

The sub label assignment is based on the confidence value from the model.

The snapshot only shows what it was at one particular time.

Your min score and threshold for fedex and usps are lower than the recommended values. You may want to increase them: https://docs.frigate.video/plus/first_model#step-4-adjust-your-object-filters-for-higher-scores

[rolf-tx]

Thank you, it was confusing with the snapshot not being consistent with the final determination.

I have increased min and threshold to 0.75.