VoryKwin / DSNet

DSNet:Double Strand Robotic Grasp Detection\Network Based on Cross Attention

PyTorch implementation of paper "DSNet:Double Strand Robotic Grasp Detection\Network Based on Cross Attention".

The article is currently under submission.

Visualization of the architecture

Datasets

Currently, both the Cornell Grasping Dataset, Jacquard Dataset and Multi Dataset are supported.

Cornell Grasping Dataset

1. Download the and extract Cornell Grasping Dataset.

Jacquard Dataset

1. Download and extract the Jacquard Dataset.

2. We use the setting in here

Training

Training is done by the main.py script.

Some basic examples:

# Train  on Cornell Dataset
python main.py   --dataset cornell

# k-fold training
python main_k_fold.py  --dataset cornell 


Trained models are saved in `output/models` by default, with the validation score appended.

## Visualize
Some basic examples:
```bash
# visualise grasp rectangles
python visualise_grasp_rectangle.py   --network your network address

# visualise heatmaps
python visualise_heatmaps.py  --network your network address

Acknowledgement

Code heavily inspired and modified from https://github.com/dougsm/ggcnn.