Can not convert Yolov4 model that trained on custom dataset to Onnx format

[Zeysthingz]

I try to run tensorrt_yolo node in Autoware Universe with trained yolov4 model on my custom dataset. This node takes onnx files then creates TensorRT engines. I convert weight and cfg files according to this repo https://github.com/Tianxiaomo/pytorch-YOLOv4 and also this repo https://github.com/linghu8812/tensorrt_inference/tree/master/Yolov4.
After converting onnx format when I run the node It gives following output and process dies.

I get same error when I convert yolov4.cfg and yolov4.weights files that already converted onnx format in Autoware Universe from scratch.

Could you provide any documentation how to convert custom darknet weights to onnx format?

[yukke42]

@wep21 @yukkysaito Could you share your conversion script?

[Zeysthingz]

Thank you a lot.

[mitsudome-r]

@wep21 ping.

[wep21]

@Zeysthingz sorry for the late reply, I prepare the script to convert darknet to onnx.
Please make sure that you change the batch size in cfg file to 1.

[Zeysthingz]

Thank you

[Zeysthingz]

@wep21 Here is my own weights ,cfg and label files. https://drive.google.com/drive/folders/1P6nNa0h1lSyy6sqUGhtO3HnG6oFwgQcN?usp=sharing