SISG-Net:Simultaneous Instance Segmentation and Grasp Detection for Robot Grasp in Clutter

The video of our experiment used in this paper can be found at: https://youtu.be/G_8USOwFXVs

The relabled OCID dataset can be found at: https://drive.google.com/file/d/1-XBstG9ur90E4X66gdG_OLVidCR56pgV/view?usp=drive_link

1. Setup anaconda environment

python=3.7
torch=1.9.0
cuda=11.1

2. Set the path to the dataset in config file.

Train

1. Download the OCID dataset at https://drive.google.com/file/d/1-XBstG9ur90E4X66gdG_OLVidCR56pgV/view?usp=drive_link.
2. Unzip the downloaded dataset, and modify the dataset_path of the config file correspondingly.
3. To train an SISG-Net on OCID dataset.

$ python train.py --gpu 0 --cfg rgb_depth