[Support]: HTTP 404 in Home Assistant Notifications

[kavejo]

Describe the problem you are having

I must be missing something pretty basic but I can't seem to be able to get the thumbnail/snapshots in the Home Assistant notifications triggered via Home Assistant Automation.

I am running Home Assistant on Docker and on the same Docker host I also run Frigate.

In Frigate I have set the "Person" detection to be an alert, while all other detections are just detections.
In Home Assistant I am checking the MQTT events and if the notification is for an alert, then HA triggers the notification.

In Frigate, for example, I have this Alert:

The notification received in the HA Companion App is missing the thumbnail (tried with the snapshot as well, with no change):

However if I press on the notification I can see the video stream correctly:

My HA Automation look like the following; I've tried to use both a custom domain (cloudflared) as well as Nabu casa.

triggers:
  - trigger: mqtt
    topic: frigate/reviews
conditions:
  - condition: template
    value_template: '{{ trigger.payload_json[''after''][''severity''] == ''alert'' }}'
actions:
  - variables:
      ha_url: MY_DOMAIN
      nabu_url: MY_NABU_CASA
  - action: notify.MY_MOBILE
    metadata: {}
    data:
      message: Persona rilevata da {{trigger.payload_json["after"]["camera"]}}
      data:
        image: >-
          {{nabu_url}}/api/frigate/notifications/{{trigger.payload_json["after"]["id"]}}/snapshot.jpg?bbox=1
        tag: '{{trigger.payload_json["after"]["id"]}}'
        when: '{{trigger.payload_json["after"]["start_time"]|int}}'
        entity_id: >-
          camera.{{trigger.payload_json["after"]["camera"] | replace("-","_") |
          lower}}
        clickAction: >-
          {{nabu_url}}/api/frigate/notifications/{{trigger.payload_json["after"]["id"]}}/snapshot.jpg?bbox=1
        push:
          sound:
            name: default
            critical: 1
            volume: 1

If I browse to the URL that is being sent from the automation, I can see the following error.

The Integration is set as per documentation:

Browsing through Media Browser however shows the image.

As such I think I must be missing something pretty basic, but I can't work out what that is.
The same code used to work when I was running HassOS and Frigate Proxy but I never got it working on this new installation.

Version

0.14.1-f4f3cfa

What browser(s) are you using?

N/A

Frigate config file

[...]
snapshots:
  enabled: true
  clean_copy: false
  timestamp: false
  bounding_box: true
  crop: false
  retain:
    default: 365
[...]

Relevant Frigate log output

N/A

Relevant go2rtc log output

N/A

FFprobe output from your camera

N/A

Frigate stats

N/A

Install method

Docker Compose

docker-compose file or Docker CLI command

HomeAssistant:
    image: "ghcr.io/home-assistant/home-assistant:stable"
    container_name: HomeAssistant
    volumes:
      - ${CONTAINER_LOCATION}/Automation/HomeAssistant:/config
      - /etc/localtime:/etc/localtime:ro
      - /run/dbus:/run/dbus:ro
    networks:
      - Automation-Net
    ports:
      - 8123:8123
    depends_on:
      MariaDB:
        condition: service_started
        restart: true
      Mosquitto:
        condition: service_started
        restart: true
    restart: always
    privileged: true

  Frigate:
    image: ghcr.io/blakeblackshear/frigate:stable
    container_name: Frigate
    privileged: true
    shm_size: "1024mb"
    devices:
      - /dev/bus/usb:/dev/bus/usb
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - ${CONTAINER_LOCATION}/NVR/Frigate/config:/config
      - ${MEDIA_LOCATION}:/media/frigate
      - type: tmpfs
        target: /tmp/cache
        tmpfs:
          size: 1000000000
    networks:
      - NVR-Net
    ports:
      - 5000:5000 # Web UI
      - 8971:8971 # Authenticated Web UI
      - 1935:1935 # RTMP
      - 8554:8554 # RTSP
      - 8555:8555/tcp # WebRTC over TCP
      - 8555:8555/udp # WebRTC over UDP
    environment:
      PLUS_API_KEY: ${PLUS_API_KEY}
      FRIGATE_MQTT_USER: ${FRIGATE_MQTT_USER}
      FRIGATE_MQTT_PASSWORD: ${FRIGATE_MQTT_PASSWORD}
      FRIGATE_RTSP_HOST_NVR: ${FRIGATE_RTSP_HOST_NVR}
      FRIGATE_RTSP_USER_NVR: ${FRIGATE_RTSP_USER_NVR}
      FRIGATE_RTSP_PASSWORD_NVR: ${FRIGATE_RTSP_PASSWORD_NVR}
#      YOLO_MODELS: yolov7-tiny-416,yolov7-640,yolov7x-640
#      USE_FP16: false
    cap_add:
      - PERFMON
    restart: always

Object Detector

Coral

Network connection

Wired

Camera make and model

Dahua

Screenshots of the Frigate UI's System metrics pages

No response

Any other information that may be helpful

No response

[NickM-27]

You are mixing concepts. The id is the id of the review item, and you are trying to use an events API. I would suggest either using the preview api to get a gif as the image on the iOS device or use the detections field which contains the IDs of each event (object)

[kavejo]

I did try to replicate what I saw on https://docs.frigate.video/guides/ha_notifications.

I read "It is generally recommended to trigger notifications based on the frigate/reviews mqtt topic. This provides the event_id(s) needed to fetch thumbnails/snapshots/clips and other useful information to customize when and where you want to receive alerts." and that was what I was trying to achieve.

Clearly I must have done something wrong.

[kavejo]

Ah, I believe I might have found the problem based on what you said @NickM-27!

I was using ["data"]["id"] which is the review id instead if ["data"]["detections"][0] which appear to be the event that triggered the detection!

Let me retry now!

[kavejo]

To confirm @NickM-27,

I have got it working by switching from "{{trigger.payload_json["after"]["id"]}}" to "{{trigger.payload_json["after"]["data"]["detections"][0]}}".
Thanks for pointing me on the right direction.

I have a follow-up question though, if you don't mind.

For an event like the following, where the review triggered has multiple detection (a car, and a person), is there a way I can get the the correct event id mapped to the person?

{
    "type": "new",
    "before": {
        "id": "1736798575.845996-b2cpc6",
        "camera": "03-Garage",
        "start_time": 1736798575.8459959,
        "end_time": null,
        "severity": "alert",
        "thumb_path": "/media/frigate/clips/review/thumb-03-Garage-1736798575.845996-b2cpc6.webp",
        "data": {
            "detections": [
                "1736798573.606801-9ab4xk",
                "1736798575.226641-tznxcn"
            ],
            "objects": [
                "car",
                "person"
            ],
            "sub_labels": [
            ],
            "zones": [
                "Stradina",
                "Marciapiede"
            ],
            "audio": [
            ]
        }
    },
    "after": {
        "id": "1736798575.845996-b2cpc6",
        "camera": "03-Garage",
        "start_time": 1736798575.8459959,
        "end_time": null,
        "severity": "alert",
        "thumb_path": "/media/frigate/clips/review/thumb-03-Garage-1736798575.845996-b2cpc6.webp",
        "data": {
            "detections": [
                "1736798573.606801-9ab4xk",
                "1736798575.226641-tznxcn"
            ],
            "objects": [
                "car",
                "person"
            ],
            "sub_labels": [
            ],
            "zones": [
                "Stradina",
                "Marciapiede"
            ],
            "audio": [
            ]
        }
    }
}

In my case I should be able to retrieve "1736798575.226641-tznxcn" and therefore the one at the position [1] instead of defaulting to the first id in the array.

Thanks!

[NickM-27]

No, there is no mapping here, in general using the preview gif API is recommended as that has a better ability to show what happened during the activity vs just choosing a single moment

[kavejo]

I will need to look into that - I am not familiar with the the preview gif API.
Thanks!