-
Notifications
You must be signed in to change notification settings - Fork 1.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
FCOS3D model conversion to onnx #1035
Comments
I'm not quietly familiar with onnx, and the release of onnx conversion is currently not in our short-term plan. Maybe you can refer to this issue: open-mmlab/mmcv#1422 |
We do have a wechat group for mmdetection3d, but mainly talk about LiDAR based 3d object detection, I'm not sure whether someone could you help. |
@ZCMax :Can I join you? I'm looking forward to joining you to discuss relevant issues. thank you. |
We have colleagues working on the deployment and model conversion for MMDet3D models. PointPillars and SECOND will have the highest priority, and then we will consider FCOS3D. So currently I will keep this issue open for community discussion, and update related changes if we make any progress. |
@Tai-Wang :Thank you for your reply and look forward to the emergence of fcos3d model transform. |
@Tai-Wang Can you estimate on when the onnx conversion for PointPillars and SECOND will be finished? |
@Thomas-Kb I can not estimate the concrete time for now. We need to finish the migration of ops from mmdet3d to mmcv first. Afterward, we will continue to work on the conversion issue. |
Hello, how can I convert FCOS3D to ONNX format? Hope to get your help, thank you |
Hi, is there any updates on how to convert FCOS3D to onnx model? thank you in advance. |
Hello, I also want to convert FCOS3D to ONNX, but it keeps failing. If you succeed, please leave me a message. thank you |
|
Hello, I also want to convert FCOS3D++ to ONNX. If there is any progress, please leave me a message. thank you!!! |
This error might occur because the custom ops for ONNX Runtime are not built correctly. Let's try to rebuild the custom ops by following the instructions in the link below |
Has anyone successfully converted the FCOS3D model to ONNX format? |
Traceback (most recent call last):
File "/home/zhanggan/mmdetection3d-master/demo/onnx-runtime.py", line 4, in
sess = rt.InferenceSession('test.onnx')
File "/home/zhanggan/anaconda3/envs/mmdetectiond3d1/lib/python3.7/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 324, in init
self._create_inference_session(providers, provider_options, disabled_optimizers)
File "/home/zhanggan/anaconda3/envs/mmdetectiond3d1/lib/python3.7/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 358, in _create_inference_session
sess = C.InferenceSession(session_options, self._model_path, True, self._read_config_from_model)
onnxruntime.capi.onnxruntime_pybind11_state.Fail: [ONNXRuntimeError] : 1 : FAIL : Load model from test.onnx failed:Fatal error: MMCVModulatedDeformConv2d is not a registered function/op
The text was updated successfully, but these errors were encountered: