ASGrasp: Generalizable Transparent Object Reconstruction and 6-DoF Grasp Detection from RGB-D Active Stereo Camera

This repository contains the source code for our paper: ASGrasp: Generalizable Transparent Object Reconstruction and 6-DoF Grasp Detection from RGB-D Active Stereo Camera

For more information, please visit our project page.

Requirements

The code has been tested with Python 3.8, PyTorch 1.9.1 and Cuda 11.1

Installation

Please refer to https://github.com/rhett-chen/graspness_implementation to install GSNet

Checkpoints

Please download the checkpoints to 'checkpoints/'

Testing

Please run

python infer_mvs_2layer_gsnet.py

Acknowledgement

Our code is based on these wonderful repos, we appreciate their great works!

• RAFT-Stereo
• IterMVS
• GraspNet
• graspness_implementation
• DREDS

Contact

If you have any questions, please open a github issue or contact us:

Jun Shi: jun7.shi@samsung.com, Yong A: yong.a@samsung.com, He Wang: hewang@pku.edu.cn