Skip to content

Latest commit

 

History

History
49 lines (36 loc) · 959 Bytes

readme.md

File metadata and controls

49 lines (36 loc) · 959 Bytes

PV-RCNN

Pytorch implementation of PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detection.

PV-RCNN

Features

  • Keypoints sampling
  • Voxel set abstraction (VSA)
  • Extended VSA
  • ROI grid pooling
  • SECOND backbone (SpMiddleFHD)
  • Predicted keypoint weighting
  • IOU-Net
  • Training code

Installation

Tested in environment:

  • CUDA 10.0
  • torch 1.0
  • Conda
  • Ubuntu 18.04
  • Python 3.6
  1. Installing Pointnet2:
git clone https://github.com/sshaoshuai/Pointnet2.PyTorch.git
cd Pointnet2.PyTorch && python setup.py install
export PYTHONPATH=$PYTHONPATH:/path/to/Pointnet2.PyTorch/
  1. Installing spconv:
git clone https://github.com/traveller59/spconv.git --recursive
cd spconv && git checkout 7342772
python setup.py bdist_wheel
cd ./dist && pip install *.whl
  1. Installing pvrcnn (this package):
python setup.py develop