We use two public datasets (six surgical tasks): JIGSAWS (suturing, knot tying, needle passing) and ROSMA (pea on a peg, post, and sleeve, wire chaser) for our experiments. Please download these two datasets using the links provided in these papers.
For JIGSAWS, we apply self-proclaimed annotation. For ROSMA, we provide high-quality annotation by two senior surgeons; the annotation will be released upon the acceptance of the paper.
We organize the pre-processed dataset into pickle format and then save them into the 'datasets/jigsaws-kinematic' folder and 'datasets/rosma-kinematic' folder.
All Python packages needed are listed in the requirments.txt file and can be installed as follows:
conda config --append channels conda-forge
conda create --name <your env name> --file requirements.txt
To run each experiment, you can run the following commands.
python CBR_module/jigsaws-su-expCBR-Feature-Fusion.py
python CBR_module/jigsaws-kt-expCBR-Feature-Fusion.py
python CBR_module/jigsaws-np-expCBR-Feature-Fusion.py
python CBR_module/rosma-pp-expCBR-Feature-Fusion.py
python CBR_module/rosma-ps-expCBR-Feature-Fusion.py
python CBR_module/rosma-wc-expCBR-Feature-Fusion.py
The results of all experiments are saved in the 'results' folder.
Example output "CBR_su_feature_fusion_scaled_new_four_views_LOSO5_results.csv" is in the 'results/jigsaws' folder.