Conversion of YOLOv7.wts(proposed custom YOLOv7) to .engine failed

Robotics & Edge Computing Jetson & Embedded Systems Jetson Nano
, ,
1

Hi,

I am deploying my own custom model i.e. PLYOLOV7 (YOLOV7 with CBAM attention module) onto Jetson Nano board. To deploy on jetson NANO , I am following below given git hub project:

github.com

GitHub - mailrocketsystems/JetsonYoloV7-TensorRT

Contribute to mailrocketsystems/JetsonYoloV7-TensorRT development by creating an account on GitHub.

Using given file of the project (JetsonYoloV7-TensorRT/buildSteps.txt at main · mailrocketsystems/JetsonYoloV7-TensorRT · GitHub) , I successfully run (conversion and testing) all standards YOLOv7 models(v7, 7x, e6e, d6, e6, w6) and test on my custom dataset which has 5 classes, on Nano board. (I first train custom PLYOLOV7 model on my custom dataset on my local GPU system then I got trained model. After that I copy that trained model on Nano board to deploy, and further I follow above given project).

During executing steps from the file buildSteps.txt for custom model PLYOLOv7, I succesfully run all the steps but when i am executing sudo ./yolov7 -s yolov7-tiny.wts yolov7-tiny.engine t (in my case : sudo ./yolov7 -s PLYOLOv7.wts PLYOLOv7.engine v7), I got following error:

error_during_conversion_from_wts-to-engine-file
error_during_conversion_from_wts-to-engine-file732×254 51.7 KB

Please help in resolving this error. If you need any details related to the topic, please let me know. Any help would be appreciated.
Thanks in advance.

3

Hi,

The error indicates an “API Usage Error”.

Do you run a default sample or custom code?
Have you checked with the API owner about this issue?

Thanks.

4

Thank you for your response.

Yes, I am running custom code, in which I added some customer layers (Convolution block attention module) in the original YOLOV7 layers or Yolov7.yaml file.

What do you mean by API usage error? I posted issue on the issue section of the git hub repository which I am following(GitHub - mailrocketsystems/JetsonYoloV7-TensorRT) but I think that is inactive now so did not receive any reply from them.

Please help in resolving issue. If you further require any information, please let me know.

5

Hi @AastaLLL

Any update?

6

Hi,

It’s recommended to check this issue with the sample owner.
As they will know more about their source.

Thanks.

7

Thank you for the reply.

Can you suggest any other way or method to run custom yolov7 code on jetson nano board?

8

Hi,

Do you have the ONNX format of the model?
If yes, you can try to convert it with trtexec tool directly.

Thanks.

10

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.