How to convert a trained model to onnx

[YueSun0609]

Thank you for your kind release of the codes, I have successfully applied it to my own data and got good performace. But may I ask how to export a trained model to onnx format? In tools/export.py, you gave an example but missing "forward_test" function. I tried to write one, but when I run torch.onnx.export() function, it meets an error like "RuntimeError: Tracer cannot infer type of [{'boxes_3d': LiDARInstance3DBoxes", could you give me some clues on how to solve this issue? Thank you so much!

[evil-master]

Thank you for your kind release of the codes, I have successfully applied it to my own data and got good performace. But may I ask how to export a trained model to onnx format? In tools/export.py, you gave an example but missing "forward_test" function. I tried to write one, but when I run torch.onnx.export() function, it meets an error like "RuntimeError: Tracer cannot infer type of [{'boxes_3d': LiDARInstance3DBoxes", could you give me some clues on how to solve this issue? Thank you so much!

请问适配到自己的数据集时，训练需要采用和nuSence一样的数据格式吗？训练也需要输入六张图像和一个lidar的点云吗（如果只有一个front图像可以有效训练吗？）

[ueke616]

I got the same error, can anyone give some solution? thanks

[zhijian-liu]

We currently do not support exporting the model to ONNX, but it is under consideration for future updates.