"require_motion" --> is my context correct?

[gniknalu]

Do I have "require_motion" in the correct context??

cameras:
  - name: "FrontWest"
    host: !secret xxxxxxxxxxxxxxxxx
    port: !secret xxxxxxxxxxxxxxxxx
    stream_format: rtsp
    username: !secret xxxxxxxxxxxxxxxxx
    password: !secret xxxxxxxxxxxxxxxxx
    path: xxxxxxxxxxxxxxxxxxxxxxxxxx
    mqtt_name: xxxxxxxxxxxxxxxxx
    publish_image: true
    zones:
      - name: DrivewayRegion
        points:
          - x: 213
            y: 158
            <............>
        labels:
          - label: person
            confidence: 0.8
            triggers_recording: true
          - label: truck
            confidence: 0.8
            triggers_recording: true
            require_motion: true
      - name: noDetectZone
        points:
          - x: 410
            y: 0
            <............>
        labels:
          - label: dog
            confidence: 1
            triggers_recording: false

I didn't want it to constantly report/record my truck.

Where I added "require_motion: true" generates this error:

voluptuous.error.MultipleInvalid: extra keys not allowed @ data['cameras'][0]['zones'][0]['labels'][1]['require_motion']
Traceback (most recent call last):
  File "viseron.py", line 8, in <module>
    from lib.config import CONFIG, NVRConfig, ViseronConfig
  File "/src/viseron/lib/config/__init__.py", line 101, in <module>
    CONFIG = VISERON_CONFIG_SCHEMA(raw_config)
  File "/usr/local/lib/python3.6/dist-packages/voluptuous/schema_builder.py", line 272, in __call__
    return self._compiled([], data)
  File "/usr/local/lib/python3.6/dist-packages/voluptuous/schema_builder.py", line 594, in validate_dict
    return base_validate(path, iteritems(data), out)
  File "/usr/local/lib/python3.6/dist-packages/voluptuous/schema_builder.py", line 432, in validate_mapping
    raise er.MultipleInvalid(errors)

[roflcoopter]

Hmm it seems correct. What does your docker command look like? Im thinking you might be running an older version

[gniknalu]

Fixed in latest dev.
It was caused by using require_motion: true on a label that wasnt being tracked under the global labels.
Can you try it out for me?

No config errors with the latest dev version.

In the zone, there is a detection (car) and the zone turns green and it records. Then it repeats. And repeats. But the car is parked.

Maybe my approach is incorrect? When a new car/truck pulls into the driveway I want a detection - different from the detection on the items that are parked in the driveway.

[roflcoopter]

require_motion only stops the recording, but not the detection. I guess this is a little bit unclear.
Am i understanding you right that you dont want the MQTT notifications either?

[gniknalu]

require_motion only stops the recording, but not the detection. I guess this is a little bit unclear.
Am i understanding you right that you dont want the MQTT notifications either?

I'm hoping to use the MQTT notification to trigger a picture sent via pushbullet. (Something I'm working on figuring out how to do using the MQTT payload.)

But when I add "require_motion" the goal is to only record when the detected object has motion. For example, two cars in the driveway - detect, fine but don't save a snapshot/video. Another car pulls into the driveway - detect yes but also record (as it is in motion or is a new detection).

[roflcoopter]

Ahh now i see the issue.
The require_motion flag only affects stopping of a recording, not starting of it.
Will fix and let you know when its up

Sorry you're having so many issues!

[gniknalu]

Sorry you're having so many issues!

I'm enjoying the journey.
No worries.

[roflcoopter]

I think i got what you need now in latest dev

I did some testing, walked infront of my camera so it started recording and then stood still, the recorder stopped when there was no motion and it did not restart until i moved.
This solves the issue you were having, yes?

[gniknalu]

I think i got what you need now in latest dev
This solves the issue you were having, yes?

Updated. It is no longer recording on the items marked "require_motion" although they're detected.
BUT it is no longer recording on any detections. Here is an example. I walked in front of two cameras. They both report something similar:

[2021-06-04 08:41:13] [DEBUG   ] [viseron.nvr.fronteast.object] - Objects: [{'label': 'person', 'confidence': 0.997, 'rel_width': 0.151, 'rel_height': 0.528, 'rel_x1': 0.604, 'rel_y1': 0.35, 'rel_x2': 0.755, 'rel_y2': 0.878}]
[2021-06-04 08:41:18] [DEBUG   ] [viseron.nvr.frontwest.object] - Objects: [{'label': 'person', 'confidence': 0.995, 'rel_width': 0.117, 'rel_height': 0.52, 'rel_x1': 0.334, 'rel_y1': 0.181, 'rel_x2': 0.451, 'rel_y2': 0.701}]

Both clearly detected me but neither recorded.

For now I'll be everting to the "latest" and commenting out the "require_motion" lines.

[roflcoopter]

😢
Can you show your full config when you ran this test?

[gniknalu]

😢
Can you show your full config when you ran this test?

cameras:
  - name: "BackEast"
    host: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    port: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    stream_format: rtsp
    username: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    password: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    path: /.................................
    mqtt_name: camera_back_east
    publish_image: true
    motion_detection:
      interval: 1
      trigger_detector: false
      threshold: 15
      alpha: 0.6
      max_timeout: 300
      mask:
        - points:
            - x: 1914
              y: 1
            - x: 320
              y: 1
            - x: 320
              y: 148
            - x: 872
              y: 33
            - x: 1914
              y: 174
    object_detection:
      interval: 1

  - name: "FrontWest"
    host: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    port: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    stream_format: rtsp
    username: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    password: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    path: /.................................
    mqtt_name: camera_front_west
    publish_image: true
    zones:
      - name: DrivewayRegion
        points:
          - x: 213
            y: 158
          - x: 765
            y: 83
          - x: 1136
            y: 121
          - x: 1264
            y: 257
          - x: 1192
            y: 445
          - x: 1274
            y: 478
          - x: 1274
            y: 543
          - x: 608
            y: 713
          - x: 384
            y: 712
          - x: 204
            y: 368
        labels:
          - label: person
            confidence: 0.8
            triggers_recording: true
          - label: dog
            confidence: 0.8
            triggers_recording: true
          - label: cat
            confidence: 0.8
            triggers_recording: true
          - label: truck
            confidence: 0.8
            triggers_recording: true
            require_motion: true
          - label: car
            confidence: 0.8
            triggers_recording: true
            require_motion: true
    motion_detection:
      interval: .5
      trigger_detector: false
      timeout: true
      max_timeout: 330
      mask:
        - points:
            - x: 153
              y: 1
            - x: 1276
              y: 1
            - x: 1276
              y: 100
            - x: 748
              y: 65
            - x: 209
              y: 105
    object_detection:
      interval: .5
  - name: "FrontEast"
    host: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    port: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    stream_format: rtsp
    username: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    password: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    path: /.................................
    mqtt_name: camera_front_east
    publish_image: true
    zones:
      - name: EDrivewayRegion
        points:
          - x: 670
            y: 1073
          - x: 649
            y: 453
          - x: 926
            y: 422
          - x: 1200
            y: 196
          - x: 1913
            y: 253
          - x: 1913
            y: 498
          - x: 1906
            y: 1073
        labels:
          - label: person
            confidence: 0.8
            triggers_recording: true
          - label: dog
            confidence: 0.8
            triggers_recording: true
          - label: cat
            confidence: 0.8
            triggers_recording: true
          - label: truck
            confidence: 0.8
            triggers_recording: true
            require_motion: true
          - label: car
            confidence: 0.8
            triggers_recording: true
            require_motion: true
    motion_detection:
      interval: .5
      trigger_detector: false
      timeout: true
      max_timeout: 330
      mask:
        - points:
            - x: 1
              y: 1
            - x: 1912
              y: 1
            - x: 1910
              y: 237
            - x: 1609
              y: 196
            - x: 1404
              y: 126
            - x: 1123
              y: 122
            - x: 922
              y: 416
            - x: 1
              y: 455
    object_detection:
      interval: .5
      
  - name: "BackWest"
    host: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    port: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    stream_format: rtsp
    username: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    password: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
    path: /.................................
    mqtt_name: camera_back_west
    publish_image: true
    motion_detection:
      interval: 1
      trigger_detector: true
      timeout: true
      max_timeout: 300
    object_detection:
      interval: .5
    logging:
      level: debug

recorder:
  extension: mkv
    filename_pattern: '%Y-%m-%d-%I%M%S'
  lookback: 10
  timeout: 10
  retain: 30
  thumbnail:
    filename_pattern: '%Y-%m-%d-%I%M%S'

# MQTT is optional
mqtt:
 broker: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
 port: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
 username: !secret xxxxxxxxxxxxxxxxxxxxxxxxx
 password: !secret xxxxxxxxxxxxxxxxxxxxxxxxx

logging:
  level: debug

[roflcoopter]

I am not seeing this behaviour.
Running with require_motion: false and require_motion: true both triggers the recorder for me.

Could you post the full log from Viseron?

[gniknalu]

Could you post the full log from Viseron?

3 of the 4 cameras above were enabled. I grabbed the latest dev image and started it. Then I walked in front of each of the 3 cameras. Verified a "detection" took place as it said "Person" on each camera. This is reflected in the logs. Two of the cameras already had a "green" zone with a detection of "truck" but there is no evidence in the logs of this happening. Then, afterwards, I checked to see if any video/snapshot was written. No files were generated.

Stopped and rm the dev image. Replaced the conf file with one that supports the "latest" image. Started the "latest" image. Repeated the test above and it did write video/snapshot files.

Here is where you can find the full log of what happened above

[roflcoopter]

I made a silly mistake...

It should work in latest dev (have you heard this one before?) 😂

It takes about 10 minutes to build so you can pull and test in a bit.

[gniknalu]

I made a silly mistake...

It should work in latest dev (have you heard this one before?) 😂

It takes about 10 minutes to build so you can pull and test in a bit.

Ran some tests. With the truck parked in the driveway - no recording. (Exactly as I would like.) When I went to the truck - the recording was triggered on detecting a "person". (As Expected.) But when I arrive, it only starts recording when I get out of the truck but doesn't "detect" and record when the truck pulls in. Well it DOES detect (I can see it turn green and say "truck") but doesn't record.

I'll run more definitive tests tomorrow especially since we'll have company pulling into the driveway.

Thank you again.

[roflcoopter]

Yeah... I need to start writing unittests but that is so much work

One thing I do in my 'day job' is automated regression testing. We should talk.

So far it's working as expected. I'll let it run through the day and will report back.

Thank you for your excellent product!

That sounds amazing! What tools are you using to do that?

[gniknalu]

That sounds amazing! What tools are you using to do that?

I'm using jMeter - currently for some .Net apps. I'm sure it could be adapted - somehow - for this app. (Possibly adding 'supportability' into this app for improved regression testing?)

[roflcoopter]

Okay i see, googled a little, is jMeter used to call APIs only?

[gniknalu]

Okay i see, googled a little, is jMeter used to call APIs only?

I use jMeter for webwalks, REST, SOAP etc. There are plugins for all types of testing.
Maybe if there was a way to test via MQTT using jMeter to listen for certain messages? Would need to publish more information though for it to be useful.