FlowNet3D: Learning Scene Flow in 3D Point Clouds (CVPR 2019)

[CVPR 2022 Oral & TPAMI 2023] Learning Optical Flow and Scene Flow with Bidirectional Camera-LiDAR Fusion

A list of point cloud scene flow papers, codes and datasets.

Code for "Learning to Segment Rigid Motions from Two Frames". CVPR 2021.

[CVPR 2021, Oral] "Weakly Supervised Learning of Rigid 3D Scene Flow"

Neural Scene Flow Prior (NeurIPS 2021 spotlight)

[CVPR 2023 Highlight 💡] Hidden Gems: 4D Radar Scene Flow Learning Using Cross-Modal Supervision

[ICRA'24] DeFlow: Decoder of Scene Flow Network in Autonomous Driving

[CVPR 2021] PV-RAFT: Point-Voxel Correlation Fields for Scene Flow Estimation of Point Clouds

The pytorch implementation of flownet3d: https://github.com/xingyul/flownet3d

Fast Neural Scene Flow (ICCV 2023)

[ECCV'24] SeFlow: A Self-Supervised Scene Flow Method in Autonomous Driving

Implementation of the FastFlow3D architecture for scene flow estimation from LiDAR point clouds in PyTorch using PyTorch Lightning.

Every Pixel Counts ++: Joint Learning of Geometry and Motion with 3D Holistic Understanding

This repo contains C++ code for sparse scene flow method.

Get 3D motion vectors / scene flow directly from Blender

Implementation of 3D SceneFlowNet: Self-Supervised 3D Scene Flow Estimation

SFAMNet: A Scene Flow Attention-based Micro-expression Network

[3DV 2024] Repository for "Multi-Body Neural Scene Flow", in International Conference on 3D Vision 2024.

Additional utility code for the Spring dataset and benchmark