Can support 3d LiDAR detection network voxelnext?

[zyxcambridge]

Request Description

https://github.com/dvlab-research/VoxelNeXt

cpu run in i9 ，

100ms

pointcloud

input 30W

Feature Use Case

No response

Issue submission checklist

• [X] The feature request or improvement must be related to OpenVINO

[zyxcambridge]

spconv

https://github.com/mit-han-lab/bevfusion/tree/main/mmdet3d/ops/spconv

[vurusovs]

@mlukasze @andrei-kochin please, take a look

[rkazants]

Hi @zyxcambridge,

Can you please elaborate more your request or issue? Is your model not converted or not inferred using OpenVINO? Try take a look at documentation how to convert PyTorch models: https://docs.openvino.ai/2023.2/openvino_docs_OV_Converter_UG_prepare_model_convert_model_Convert_Model_From_PyTorch.html

Best regards, Roman

[zyxcambridge]

pth is here https://drive.google.com/file/d/1IV7e7G9X-61KXSjMGtQo579pzDNbhwvf/view?usp=share_link

infrence code is here https://github.com/dvlab-research/VoxelNeXt/blob/master/tools/demo.py

unsupport os is called spconv https://github.com/mit-han-lab/bevfusion/tree/main/mmdet3d/ops/spconv ;

the network is smilar to pointpillars ,but now openvino not support any of op in voxelnext ?

[rkazants]

pth is here https://drive.google.com/file/d/1IV7e7G9X-61KXSjMGtQo579pzDNbhwvf/view?usp=share_link

infrence code is here https://github.com/dvlab-research/VoxelNeXt/blob/master/tools/demo.py

unsupport os is called spconv https://github.com/mit-han-lab/bevfusion/tree/main/mmdet3d/ops/spconv ;

the network is smilar to pointpillars ,but now openvino not support any of op in voxelnext ?

Hi @zyxcambridge,

Can you please share your logs you received during convert_model to see a concrete list of unsupported ops?

Best regards, Roman