-------------------------------18 March 2022------------------------------------

1、PointNet++

Weights_trained_on_ShapeNet shows the PointNet++ network we used to gain pretrained models.
The checkpoint floder contains the weights we trained on ShapeNet Dataset.
Please refer the PointNet++ article for more details.

2、mmdection3D

I found mmdetection3D contain some 3D compoment segmentation backbone, but the dataset they used are ScanNet, SUNRGB-D, Waymo, nuScenes, Lyft, and KITTI.

3、3D Part Segmentation on ShapeNet

I found a leadboard : 3D Part Segmentation on ShapeNet dataset.

Model	Backbone	Url	Journal
KPConv	CNN-based	https://github.com/HuguesTHOMAS/KPConv	ICCV 2019
GDANet	CNN-based	https://github.com/mutianxu/GDANet	AAAI2021
Spherical Kernel	Graph-Based	https://github.com/EnyaHermite/SPH3D-GCN	CVPR2019
PointCNN	CNN-Based	https://github.com/yangyanli/PointCNN	NeurIPS 2018
DRNet	CNN-Based	https://github.com/ShiQiu0419/DRNet	WACV 2021
ConvPoint	CNN-Based	https://github.com/aboulch/ConvPoint/tree/master/examples/shapenet	Computers & Graphics 2019
SpiderCNN	CNN-Based	https://github.com/xyf513/SpiderCNN	ECCV 2018
3DGCN	Graph-Based	https://github.com/zhihao-lin/3dgcn/tree/master/segmentation	CVPR2020
Splatnet	CNN-Based	https://github.com/NVlabs/splatnet	CVPR2018
Point Transformer	Transformer-Based	https://github.com/engelnico/point-transformer	IEEE Access 2021
CurveNet	CNN-Based	https://github.com/tiangexiang/CurveNet#point-cloud-part-segmentation	ICCV 2021
PVT	Transformer-Based	https://github.com/HaochengWan/PVT	2021
PVCNN volumetric	CNN-Based	https://github.com/Yochengliu/Relation-Shape-CNN	CVPR 2019
DGCNN	Graph-Based	https://github.com/WangYueFt/dgcnn	2018
PCNN	CNN-Based	https://github.com/matanatz/pcnn	SIGGRAPH 2018
SO-Net	Graph-Based	https://github.com/lijx10/SO-Net	CVPR 2018
Pointnet++-SSG	MLP-Based	https://github.com/charlesq34/pointnet2	NIPS 2017
PointNet++-MSG	MLP-Based	https://github.com/charlesq34/pointnet2	NIPS 2017
Pointnet	MLP-Based	https://github.com/charlesq34/pointnet	CVPR2017

-------------------------------25 March 2022------------------------------------

4、Lidar SLAM

We published the code we used to gain lidar point clouds in Lidar_SLAM.
The code borrows a lot from Open3d

-------------------------------26 March 2022------------------------------------

5、Registration

Teaser
ICP

Acknowledgement

Our code borrows a lot from:
PointNet++
PointNet
Open3d
Teaser

The code is constantly being updated