[Config Support]: No detection reported

[Jimmi61]

Describe the problem you are having

I'm testing the configuration of Frigate and so far it's everything good, no errors detected, except for the fact that doesn't seems to detect anything. I'm moving in front of the camera but I can't see any reaction. What am I missing?

I also noted something strange. I read somewhere that could be better to reduce the FPS of the camera do have FRigate working better, but if I reduce the FPS from the default 25 to 15 I start having the ffmpeg process start crashing.

Version

0.16.1-e664cb2

Frigate config file

mqtt:
  host: hs.local
  user: "{FRIGATE_MQTT_USER}"
  password: "{FRIGATE_MQTT_PASSWORD}"

ffmpeg:
  global_args: -hide_banner -loglevel warning -threads 4
  hwaccel_args: preset-vaapi
  input_args:
    - preset-vaapi
    - preset-rtsp-generic
    # - +igndts
  output_args:
    detect: -threads 4 -f rawvideo -pix_fmt yuv420p
    record: preset-record-generic

# snapshots:
#   enabled: true
#   timestamp: true
#   bounding_box: false
#   crop: false
#   retain:
#     default: 10
#     objects:
#       person: 15

timestamp_style:
  position: tl
  format: '%H:%M:%S'
  color:
    red: 255
    green: 255
    blue: 255
  thickness: 2

objects:
  track:
    - person
    - cat
    - dog
    - car
#    - bicycle
  filters:
    person:
      min_area: 5000
      max_area: 100000
      min_score: 0.5
      threshold: 0.7
      mask: 0,0,1000,0,1000,200,0,200
    car:
      mask:
        - 0,0,1000,0,1000,200,0,200

# record:
#   enabled: true
#   retain:
#     days: 0
#   alerts:
#     retain:
#       days: 10
#     pre_capture: 15
#     post_capture: 15
#   detections:
#     retain:
#       days: 10
#     pre_capture: 15
#     post_capture: 15

detect:
  enabled: true
  stationary:
    interval: 1
    threshold: 20

go2rtc:
  streams:
    CCTV_1:
      - rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@10.0.10.3/cam/realmonitor?channel=1&subtype=0

cameras:
  dahua_0:
    enabled: true
    ffmpeg:
      output_args:
        record: preset-record-generic-audio-copy
      inputs:
        - path: rtsp://127.0.0.1:8554/CCTV_1
          input_args: preset-rtsp-restream
          roles:
            - detect
            - record
    objects:
      filters:
        person:
          mask: 0,0,1,0,1,1,0,1
        car:
          mask: 0,0,1,0,1,1,0,1

detectors:
  hailo8l:
    type: hailo8l
    device: PCIe

model:
  width: 320
  height: 320
  input_tensor: nhwc
  input_pixel_format: rgb
  input_dtype: int
  model_type: yolo-generic

version: 0.16-0

Relevant Frigate log output

info | 2025-09-10 13:38:56 | startup                 | Preparing Frigate...
info | 2025-09-10 13:38:56 | startup                 | Starting Frigate...
info | 2025-09-10 13:38:58 | frigate.util.config     | Checking if frigate config needs migration...
info | 2025-09-10 13:38:58 | frigate.util.config     | frigate config does not need migration...
info | 2025-09-10 13:39:00 | frigate.app             | Starting Frigate (0.16.1-e664cb2)
info | 2025-09-10 13:39:00 | peewee_migrate.logs     | Starting migrations
info | 2025-09-10 13:39:00 | peewee_migrate.logs     | There is nothing to migrate
info | 2025-09-10 13:39:00 | frigate.app             | Recording process started: 382
info | 2025-09-10 13:39:00 | frigate.app             | Review process started: 391
info | 2025-09-10 13:39:00 | frigate.app             | go2rtc process pid: 121
info | 2025-09-10 13:39:00 | detector.hailo8l        | Starting detection process: 407
info | 2025-09-10 13:39:00 | frigate.app             | Output process started: 426
info | 2025-09-10 13:39:00 | frigate.app             | Camera processor started for dahua_0: 460
info | 2025-09-10 13:39:00 | frigate.app             | Capture process started for dahua_0: 464
info | 2025-09-10 13:39:01 | frigate.api.fastapi_app | Starting FastAPI app
info | 2025-09-10 13:39:01 | frigate.api.fastapi_app | FastAPI started

Relevant go2rtc log output

info | 2025-09-10 13:38:56 | startup | Preparing new go2rtc config...
info | 2025-09-10 13:38:56 | startup | Starting go2rtc...
info | 2025-09-10 13:38:56 | startup | go2rtc platform=linux/amd64 revision=fa580c5 version=1.9.9
info | 2025-09-10 13:38:56 | startup | config path=/dev/shm/go2rtc.yaml
info | 2025-09-10 13:38:56 | rtsp    | listen addr=:8554
info | 2025-09-10 13:38:56 | api     | listen addr=:1984
info | 2025-09-10 13:38:56 | webrtc  | listen addr=:8555
info | 2025-09-10 13:39:06 | startup | Starting go2rtc healthcheck service...

Frigate stats

No response

Operating system

Debian

Install method

Docker Compose

docker-compose file or Docker CLI command

services:
  frigate:
    container_name: frigate
    image: ghcr.io/blakeblackshear/frigate:stable
    shm_size: "512mb" # calculate on your own
    stop_grace_period: 30s # allow enough time to shut down the various services
    restart: unless-stopped
    privileged: true
    env_file:
      - ./common.env
      - ./secret.env
    ports:
      - "5000:5000"
      - "1935:1935"
      - "8554:8554"
      - "8555:8555/tcp" # WebRTC over tcp
      - "8555:8555/udp" # WebRTC over udp
    devices:
      - /dev/bus/usb:/dev/bus/usb
      - /dev/hailo0:/dev/hailo0
      - /dev/video0:/dev/video0
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - ./frigate/config:/config
      - ./frigate/storage:/media/frigate
      - /tmp/cache:/tmp/cache

Object Detector

Other

Screenshots of the Frigate UI's System metrics pages

Any other information that may be helpful

Using a minipc EQ14 with a Hailo 8 PCIe card. I installed Debian Trixie and run Frigate in a Docker container.

[inkeep-ai-assistant[bot]]

Based on your configuration, I can see several issues that are preventing detection:

Main Detection Issues

1. Detection is globally disabled: Your global detect configuration has enabled: false (1). This needs to be enabled: true for object detection to work.

2. Overly restrictive masks: Your object filters are using masks that appear to cover almost the entire frame:

   • Person mask: 0,0,1,0,1,1,0,1
   • Car mask: 0,0,1,0,1,1,0,1

These normalized coordinates (0-1 range) create tiny masks that would filter out nearly all detections (2).

Configuration Fixes

Update your configuration:

detect:
  enabled: true  # Change this from false to true
  stationary:
    interval: 1
    threshold: 20

(1)

For the masks, if you want to exclude areas like the top portion of the frame, use proper coordinates. The mask format uses normalized coordinates where the frame dimensions are 0.0 to 1.0 (2).

Motion Detection

Motion detection is required for object detection (3). Since you don't have a motion section configured, it should default to enabled, but verify motion is being detected in the debug view (Settings → Debug with Motion Boxes enabled) (2).

Hardware Configuration

Your Hailo8L detector configuration looks correct for your hardware setup (4):

detectors:
  hailo8l:
    type: hailo8l
    device: PCIe

(4)

After making these changes, restart Frigate and check if detections start working. The debug view will help you see if motion is being detected and if the masks are properly configured.

📚 Sources:

1. Full Reference Config
2. Masks
3. Full Reference Config
4. Supported Hardware

---

Have another question? Just tag @inkeep.

[Jimmi61]

detect was already set to true, the masks instead were wrong, my bad. I removed the masks, also those for objects, and restarted but again nothing happens, no detection.

If I go in the debug view I see the motion correctly detected, the motion boxes appears, also the timestamp, but nothing of them are shown in the live view and the count of detection or motion remains 0.

[hawkeye217]

I note a few issues with your config:

1. input_args should not be a list. You should use a single preset: https://docs.frigate.video/configuration/ffmpeg_presets#input-args-presets

2. Detection is not functioning because you are missing parameters for your model configuration. See the official docs on configuring the Hailo detector, specifically labelmap_path: https://docs.frigate.video/configuration/object_detectors#configuration

[Jimmi61]

thanks @hawkeye217 for pointing out. I changed both mistakes, here is the last config.yaml:

mqtt:
  host: hs.local
  user: "{FRIGATE_MQTT_USER}"
  password: "{FRIGATE_MQTT_PASSWORD}"

ffmpeg:
  global_args: -hide_banner -loglevel warning -threads 4
  hwaccel_args: preset-vaapi
  input_args: preset-rtsp-generic
  output_args:
    detect: -threads 4 -f rawvideo -pix_fmt yuv420p
    record: preset-record-generic

timestamp_style:
  position: tl
  format: '%H:%M:%S'
  color:
    red: 255
    green: 255
    blue: 255
  thickness: 2

objects:
  track:
    - person
    - cat
    - dog
    - car

detect:
  enabled: true
  stationary:
    interval: 1
    threshold: 20

go2rtc:
  streams:
    CCTV_1:
      - rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@10.0.10.3/cam/realmonitor?channel=1&subtype=0

cameras:
  dahua_0:
    enabled: true
    ffmpeg:
      output_args:
        record: preset-record-generic-audio-copy
      inputs:
        - path: rtsp://127.0.0.1:8554/CCTV_1
          input_args: preset-rtsp-restream
          roles:
            - detect
            - record

detectors:
  hailo8l:
    type: hailo8l
    device: PCIe

model:
  width: 320
  height: 320
  input_tensor: nhwc
  input_pixel_format: rgb
  input_dtype: int
  model_type: yolo-generic
  labelmap_path: /labelmap/coco-80.txt

version: 0.16-0

One thing I did not understood before: the camera was upside down (ceiling camera laying on the table) and the person was not detected. A bit weird for me but this is the way it works, good to understand.

I turned the camera and now, when it catch me, the whole frame is surrounded by a red line and the icon of a person appears on top left, not exactly the behavior I expected. In debug view the motion boxes correctly appears. And still in the review tab there are 0 detection and no motion detection.

I also try to add specific configuration such as:

motion:
  enabled: True

review:
  alerts:
    enabled: True
    labels:
      - car
      - person
      - cat
      - dog
  detections:
    enabled: True
    labels:
      - car
      - person
      - cat
      - dog

but it made no difference.

BTW in debug view I can also see the timestamp, that is not visible in Live view, is it normal?

[NickM-27]

You don't have recordings enabled so you won't get alerts or detections

[Jimmi61]

Thanks @NickM-27, too many things were not clear for me. I'm getting closer but still something is strange, if not wrong.

I enabled recodings and snapshopts and now they are working. However in the review I see only Alarms, zero Detection and only one Motion out of 13 recorded alarms. if I go in the Explore view the title says 'Person 43 tracked objects'.

I saw that the GPU usage never moved from 0.0%. I had the doubt that the 'preset-vaapi' I choose was correct, and I changed with 'preset-intel-qsv-h264'. Now the GPU usage says 0.01%, but still I'm not sure about the choice. My mini PC has Intel Twin Lake N150 and UHD Graphics 24EUs GPU.
In Docker I passed the device '/dev/dri/renderD128' that exists and should be the correct one.

Now my configuration is:

mqtt:
  host: hs.local
  user: "{FRIGATE_MQTT_USER}"
  password: "{FRIGATE_MQTT_PASSWORD}"

ffmpeg:
  global_args: -hide_banner -loglevel warning -threads 4
  hwaccel_args: preset-intel-qsv-h264
  input_args: preset-rtsp-generic
  output_args:
    detect: -threads 4 -f rawvideo -pix_fmt yuv420p
    record: preset-record-generic

snapshots:
  enabled: true
  timestamp: true
  bounding_box: true
  crop: false
  retain:
    default: 10
    objects:
      person: 15

timestamp_style:
  position: tl
  format: '%H:%M:%S'
  color:
    red: 255
    green: 255
    blue: 255
  thickness: 4

objects:
  track:
    - person
    - cat
    - dog
    - car

review:
  alerts:
    enabled: True
    labels:
      - car
      - person
      - cat
      - dog
  detections:
    enabled: True
    labels:
      - car
      - person
      - cat
      - dog

motion:
  enabled: True

record:
  enabled: true
  retain:
    days: 0
  alerts:
    retain:
      days: 10
    pre_capture: 15
    post_capture: 15
  detections:
    retain:
      days: 10
    pre_capture: 15
    post_capture: 15

detect:
  enabled: true
  stationary:
    interval: 1
    threshold: 20

go2rtc:
  streams:
    CCTV_1:
      - "ffmpeg:rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@10.0.10.3/cam/realmonitor?channel=1&subtype=0#video=h264#rotate=180"

cameras:
  dahua_0:
    enabled: true
    ffmpeg:
      output_args:
        record: preset-record-generic-audio-copy
      inputs:
        - path: rtsp://127.0.0.1:8554/CCTV_1
          input_args: preset-rtsp-restream
          roles:
            - detect
            - record
 
detectors:
  hailo8l:
    type: hailo8l
    device: PCIe

model:
  width: 320
  height: 320
  input_tensor: nhwc
  input_pixel_format: rgb
  input_dtype: int
  model_type: yolo-generic
  labelmap_path: /labelmap/coco-80.txt

version: 0.16-0

[NickM-27]

That comes down to your config, you've made it so that alerts and detections are the same thing, so naturally they're all alerts. You should setup required zones.

Motion only shows one view for each camera and you can scrub the timeline

There's an Intel bug that shows GPU as 0%

[Jimmi61]

Thanks @NickM-27 , this clarified to me another concept I couldn't grab before. I thought Alerts and Detection were two separate entities but I understand they are not, Detection is a sort of level 0 while Alerts is a level 1. Also Motion now is clear.

Regarding the GPU I found this and this. I understand that there are non-free drivers that may solve the issue and in future we may have them included in Frigate container.

[NickM-27]

Non free drivers are already included, to be clear