System.NullReferenceException

[fireport]

I'd installed version 1.67 of this lovely tools on my windows 10 Blue Iris Machine but i'm unable to use it.
In the log i only sse this error

51.023]: Starting analysis of C:\ai_images\laterale.20200905_141251250.jpg
[05.09.2020, 14:12:51.042]: (1/6) Uploading image to DeepQuestAI Server
[05.09.2020, 14:12:51.064]: (2/6) Waiting for results
[05.09.2020, 14:12:51.080]: (3/6) Processing results:
[05.09.2020, 14:12:51.097]: System.NullReferenceException | Riferimento a un oggetto non impostato su un'istanza di oggetto. (code: -2147467261 )
[05.09.2020, 14:12:51.116]: ERROR: Processing the following image 'C:\ai_images\laterale.20200905_141251250.jpg' failed. Failure in AI Tool processing the image.

Deepstack is installed on a linux machime using docker and, with demo code, works like a charms.
Can you help me ?

[VorlonCD]

Try my fork of the AI Tool: https://github.com/VorlonCD/bi-aidetection/releases

[feene]

I am getting the same code. code: -2147467261 - running deepstack noavx

Log from Deepstack
[GIN] 2020/10/03 - 02:30:50 | 200 | 153.12µs | 172.16.1.211 | POST /v1/vision/detection
[GIN] 2020/10/03 - 02:31:53 | 200 | 23.891µs | 172.16.1.211 | POST /v1/vision/detection

From AI
[02.10.2020, 20:29:33.655]: Starting analysis of F:\AI-Snaps\Garage1.20201002_202933533.jpg
[02.10.2020, 20:29:33.655]: System.IO.IOException | The process cannot access the file 'F:\AI-Snaps\Garage1.20201002_202933533.jpg' because it is being used by another process. (code: -2147024864 )
[02.10.2020, 20:29:33.655]: Could not access file - will retry after 10 ms delay
[02.10.2020, 20:29:33.667]: Retrying image processing - retry 1
[02.10.2020, 20:29:33.669]: System.IO.IOException | The process cannot access the file 'F:\AI-Snaps\Garage1.20201002_202933533.jpg' because it is being used by another process. (code: -2147024864 )
[02.10.2020, 20:29:33.669]: Could not access file - will retry after 20 ms delay
[02.10.2020, 20:29:33.691]: Retrying image processing - retry 2
[02.10.2020, 20:29:33.691]: System.IO.IOException | The process cannot access the file 'F:\AI-Snaps\Garage1.20201002_202933533.jpg' because it is being used by another process. (code: -2147024864 )
[02.10.2020, 20:29:33.693]: Could not access file - will retry after 30 ms delay
[02.10.2020, 20:29:33.725]: Retrying image processing - retry 3
[02.10.2020, 20:29:33.725]: (1/6) Uploading image to DeepQuestAI Server
[02.10.2020, 20:29:33.733]: (2/6) Waiting for results
[02.10.2020, 20:29:33.735]: (3/6) Processing results:
[02.10.2020, 20:29:33.735]: System.NullReferenceException | Object reference not set to an instance of an object. (code: -2147467261 )
[02.10.2020, 20:29:33.735]: ERROR: Processing the following image 'F:\AI-Snaps\Garage1.20201002_202933533.jpg' failed. Failure in AI Tool processing the image.

[feene]

-e VISION-SCENE=True -e VISION-DETECTION=True

Add to your Docker Start up. Was Missing the Detection Piece.

[mazcoder]

I’ve been having the same problem/error code with my set up. Turns out I had a “true” instead of a “True” in the VISION-DETECTION Env in portainer.

[Thermaltech]

This error seems to be due to the lack of a null object trap on response.predictions before line 246 in Shell.cs.

Although I have not spent more than 5 mins looking at the code and deepstack API examples, it seems that if there are no objects detected in a picture, the response would be NULL or an empty JSON object, which the code does not account for prior to the FOR / EACH loop. I may be wrong, but hopefully it points in the right direction. The deepstack API does not seem to be well documented / published, only examples given, so hard to say for sure.

After a little more testing, and adding the VISION-SCENE=True environment variable to docker, the error seems to have gone away, but I think the reason, is that when VISION-SCENE is enabled, you never get a empty object returned, as there is always some type of scene in front of the camera, but the object is discarded / ignored as it is not one of the AI Tools standard relavent object types...

[kozoke]

Same issue.
I added:
-e VISION-SCENE=True
I had this already: -e VISION-DETECTION=True

However, I had true instead of a capitalized, True

Good luck! I'm up and running!