Axis Communications
Model porting issue #1191
Replies: 6 comments 13 replies
-
|
This automatically generated reply acts as a friendly reminder. Answers to your questions will most often come from the community, from developers like yourself. You will, from time to time, find that Axis employees answers some of the questions, but this is not a guarantee. Think of the discussion forum as a complement to other support channels, not a replacement to any of them. If your question remains unanswered for a period of time, please revisit it to see whether it can be improved by following the guidelines listed in Axis support guidelines. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @kavitaRasal , These guides may help you reach the root cause: Please upload the camera server reports for both models. Checking the server report |
Beta Was this translation helpful? Give feedback.
-
|
Uploading Axis_SR_20251204_101401_E827250BBB11.zip… |
Beta Was this translation helpful? Give feedback.
-
|
Link not working. |
Beta Was this translation helpful? Give feedback.
-
|
@kavitaRasal for me it looks like the model use a unsporting instruction TRANSPOSE_CONV [67] https://developer.axis.com/computer-vision/computer-vision-on-device/optimization-tips/ trans_conv only works on artpec-8 |
Beta Was this translation helpful? Give feedback.
-
|
Hi @kavitaRasal, Since you are using another model than the one supplied in the example, maybe the parsing of the output tensors is the problem 🤔 The parsing is from line 188 to line 211 in vdo_larod.c: 
Maybe you can remove those rows and see if the example works better 😉 |
Beta Was this translation helpful? Give feedback.
-
|
after removing postprocessing crash is not happening. looking into post processing code
|
Beta Was this translation helpful? Give feedback.
-
|
I was able to fix the crash after checking post processing logic. But I get number of object always 0. Can I get camera with some objects moving , so that I can verify that I do get some objects or not. |
Beta Was this translation helpful? Give feedback.
-
|
Vivek,
AXIS P1475-LE Bullet Camera
I am using this camera to test .
I was able to fix the crash after checking post processing logic. But I get number of object always 0. Can I get camera with some objects moving , so that I can verify that I do get some objects or not.
Thanks
Kavita
|
Beta Was this translation helpful? Give feedback.
-
|
@kavitaRasal which model did you use ? Should not like in yolo the output tensor full of data because it's the raw tensor output and you would do checking score/confidence etc. eg filter. So when you say 0 objects did you mean empty output tensors or something else ? |
Beta Was this translation helpful? Give feedback.
-
|
@kavitaRasal I suppose you use virtual loan cameras. When you can detect static objects you can try overlay and image of the object via Webui. because I think this runs before vdo gets the frame. |
Beta Was this translation helpful? Give feedback.
-
Hi @kavitaRasal , |
Beta Was this translation helpful? Give feedback.
-
|
May be some vehicle or person moving in front of camera would be fine |
Beta Was this translation helpful? Give feedback.
-
|
That will not be possible in the AXIS Virtual Loan Tool. I will request the team to keep this camera in front of the TV 😉 If you don't have a physical camera, then you can try this: |
Beta Was this translation helpful? Give feedback.
-
|
Hi, Below are the error I am getting while loading Successfuly loaded for CPU: 2025-12-16T04:19:52.718+00:00 axis-b8a44ff11728 [ INFO ] vdo_larod_new[91243]: 12 Starting /usr/local/packages/vdo_larod_new/vdo_larod_new Failed for ARTPEC9: 2025-12-16T04:24:58.657+00:00 axis-b8a44ff11728 [ INFO ] vdo_larod_new[92776]: 12 Starting /usr/local/packages/vdo_larod_new/vdo_larod_new |
Beta Was this translation helpful? Give feedback.
-
|
Hi @kavitaRasal , Which may help you to identify the issue? |
Beta Was this translation helpful? Give feedback.
-
|
Models here do not have output more than 4. This does not answer my question, whether axis support 12 outputs? |
Beta Was this translation helpful? Give feedback.
-
|
Hi @kavitaRasal, it should not have anything to do with the number of output tensors, as you suggest. Will get back to you about the type of logs required from your end to figure out what is going wrong. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @kavitaRasal , |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
I am trying to port tensorflow models to tflite.
While conversion Transpose layer gets added.
When I try to use this model with VDC-larod sample application. App crashes while processing first frame.
Here are app logs.
Can you please suggest how to fix this? Are we following correct steps?
2025-11-20T10:56:09.457+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Starting new Version 4:27pm /usr/local/packages/vdo_larod/vdo_larod
2025-11-20T10:56:10.007+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Setting up larod connection with chip a9-dlpu-tflite and model file /usr/local/packages/vdo_larod/model/model.tflite
2025-11-20T10:56:10.008+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Loading the model... This might take up to 5 minutes depending on your device model.
2025-11-20T10:56:11.239+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Model loaded successfully
2025-11-20T10:56:11.240+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Detected model format RGB and input resolution 448x448
2025-11-20T10:56:11.240+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Created mmaped model output 0 with size 3136
2025-11-20T10:56:11.240+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Created mmaped model output 1 with size 196
2025-11-20T10:56:11.241+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Created mmaped model output 2 with size 3136
2025-11-20T10:56:11.241+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Created mmaped model output 3 with size 784
2025-11-20T10:56:11.241+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Created mmaped model output 4 with size 3136
2025-11-20T10:56:11.241+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Created mmaped model output 5 with size 12544
2025-11-20T10:56:11.241+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Created mmaped model output 6 with size 784
2025-11-20T10:56:11.241+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Created mmaped model output 7 with size 784
2025-11-20T10:56:11.241+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Created mmaped model output 8 with size 12544
2025-11-20T10:56:11.251+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: choose_stream_resolution: We select stream w/h=1024 x 576 with format rgb interleaved based on VDO channel info.
2025-11-20T10:56:11.251+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Dump of vdo stream settings map =====
2025-11-20T10:56:11.251+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: 'buffer.count'-----: <uint32 2>
2025-11-20T10:56:11.251+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: 'dynamic.framerate':
2025-11-20T10:56:11.251+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: 'format'-----------: <uint32 8>
2025-11-20T10:56:11.251+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: 'framerate'--------: <30.0>
2025-11-20T10:56:11.251+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: 'height'-----------: <uint32 576>
2025-11-20T10:56:11.251+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: 'input'------------: <uint32 1>
2025-11-20T10:56:11.251+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: 'socket.blocking'--:
2025-11-20T10:56:11.251+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: 'width'------------: <uint32 1024>
2025-11-20T10:56:11.261+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Use preprocessing with input format rgb-interleaved and output format rgb-interleaved
2025-11-20T10:56:11.263+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Calculate crop image
2025-11-20T10:56:11.263+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Crop input image X=224 Y=0 (576 x 576)
2025-11-20T10:56:11.263+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Start fetching video frames from VDO
2025-11-20T10:56:11.833+00:00 axis-b8a44ff141cb [ INFO ] vdo_larod[48853]: Ran inference for 337 ms
axis-e82725122bf1:~$ tail -f /var/log/error.log
2025-11-24T12:28:46.771+00:00 axis-e82725122bf1 [ ERR ] larod[290478]: Operator TRANSPOSE_CONV [67] is not supported by armnn_delegate.
2025-11-24T12:28:48.122+00:00 axis-e82725122bf1 [ ERR ] kernel: vdo_larod: vdo_larod: potentially unexpected fatal signal 11.
2025-11-24T12:50:37.254+00:00 axis-e82725122bf1 [ ERR ] larod[290478]: TfLiteArmnnDelegate: Unexpected number of inputs (4 != 3) in node #257
2025-11-24T12:50:37.275+00:00 axis-e82725122bf1 [ ERR ] larod[290478]: TfLiteArmnnDelegate: Unexpected number of inputs (4 != 3) in node #304
2025-11-24T12:50:37.294+00:00 axis-e82725122bf1 [ ERR ] larod[290478]: Operator TRANSPOSE_CONV [67] is not supported by armnn_delegate.
2025-11-24T12:50:38.710+00:00 axis-e82725122bf1 [ ERR ] kernel: vdo_larod: vdo_larod: potentially unexpected fatal signal 11.
2025-11-25T04:47:11.960+00:00 axis-e82725122bf1 [ ERR ] larod[290478]: TfLiteArmnnDelegate: Unexpected number of inputs (4 != 3) in node #257
2025-11-25T04:47:11.979+00:00 axis-e82725122bf1 [ ERR ] larod[290478]: TfLiteArmnnDelegate: Unexpected number of inputs (4 != 3) in node #304
2025-11-25T04:47:11.998+00:00 axis-e82725122bf1 [ ERR ] larod[290478]: Operator TRANSPOSE_CONV [67] is not supported by armnn_delegate.
2025-11-25T04:47:13.286+00:00 axis-e82725122bf1 [ ERR ] kernel: vdo_larod: vdo_larod: potentially unexpected fatal signal 11.
What could be the reason for this crash
Beta Was this translation helpful? Give feedback.
All reactions