I trained an object detection model using jetson-inference (not using docker container). It successfully trained 40 epochs but when converting it to ONNX I get a Segmenation fault (core dumped) error:
Hi @akashsivapalanspam, it looks like it encountered the error before it actually exported the ONNX file. Have you been able to export the ONNX of other models (or classification models)?
Can you try exporting a different checkpoint from your training with the --input and --output arguments to onnx_export.py?
You might also need to set --labels when manually selecting the checkpoint.