Skip to content

Latest commit

 

History

History
executable file
·
15 lines (12 loc) · 1.16 KB

README.md

File metadata and controls

executable file
·
15 lines (12 loc) · 1.16 KB

USEEK: Unsupervised SE(3)-Equivariant 3D Keypoints for Generalizable Manipulation (ICRA2023)

Project website: https://sites.google.com/view/useek. The paper is available at https://arxiv.org/abs/2209.13864.

A map of the repository

  • 1_train_merger.py trains the teacher network, following the procedure in Skeleton Merger with minor adaptations.
  • 2_prepare_segment.py prepares the segmentation pseudo labels for the student network.
  • 3_train_useek.py trains the student network, which is adapted from the segmentation network of SPRIN.
  • 4_predictor_keypointnet.py and 5_eval_keypointnet.py test the student network on the KeypointNet dataset with SE(3) transformations.

Dataset

The ShapeNetCore.v2 dataset used for training is available from the Point Cloud Datasets repository.

Pre-trained models

The pre-trained models for both the teacher and student networks are available at Google Drive.