Home

Awesome

Multi-Scale Bidirectional Recurrent Network with Hybrid Correlation for Point Cloud Based Scene Flow Estimation

Wencan Cheng and Jong Hwan Ko

IEEE International Conference on Computer Vision (ICCV), 2023

Prerequisities

Our model is trained and tested under:

Please follow this repo or the instructions below for compiling the furthest point sampling, grouping and gathering operation for PyTorch.

cd pointnet2
python setup.py install
cd ../

Data preprocess

We adopt the equivalent preprocessing steps in HPLFlowNet and PointPWCNet.

We copy the preprocessing instructions here for your convinience.

python3 data_preprocess/process_flyingthings3d_subset.py --raw_data_path RAW_DATA_PATH --save_path SAVE_PATH/FlyingThings3D_subset_processed_35m --only_save_near_pts
python3 data_preprocess/process_kitti.py RAW_DATA_PATH SAVE_PATH/KITTI_processed_occ_final

Evaluation

Set data_root in the configuration file to SAVE_PATH in the data preprocess section before evaluation.

We provide pretrained model in pretrain_weights.

Please run the following instrcutions for evaluating.

python3 evaluate.py config_evaluate.yaml

Train

If you need a newly trained model, please set data_root in the configuration file to SAVE_PATH in the data preprocess section before evaluation at the first. Then excute following instructions.

python3 train_msbrn.py config_train.yaml

Acknowledgement

We thank repo for the corase-to-fine framework.