Directional Distance Field for Modeling the Difference between 3D Point Clouds [OpenReview]

Here we provide the examples for three tasks, including shape reconstruction, rigid registration and scene flow estimation.

Requirements

pytorch             #1.10.0+cu111
pytorch3d           #0.6.2
open3d
trimesh
point-cloud-utils

Shape Reconstruction

python exp_shape_reconstruction.py

The reconstructed point cloud and trained model is saved at example/shape_reconstruction.

Rigid Registration

run python exp_registration.py to get the RE and TE.

(Note the rotation matrix here is the transpose of the GT rotation matrix.)

Scene Flow Estimation

run python exp_scene_flow.py to get the EPE, ACC-0.05, ACC-0.1, and Outliers.