Home

Awesome

AdaptiveSAM

This repository contains the code for AdaptiveSAM: Towards Efficient Tuning of SAM for Surgical Scene Segmentation

Environment File

Create a new conda environment with the config file given in the repository as follows:

conda env create --file=biastuning_env.yaml
conda activate biastuning_env

General file descriptions

Link to model checkpoints

GDrive

Example Usage for Training

python driver_scratchpad.py --model_config model_biastuning.yml --data_config config_cholec8k.yml --save_path "./temp.pth"

Example Usage for Evaluation

cd eval/endovis

python generate_predictions.py --model_config config_model_test.yml --data_config config_endovis_test.yml --data_folder <path to image folder> --gt_path <path to ground truth images folder> --save_path "./temp_results" --pretrained_path <path to model>

Citation

@misc{paranjape2023adaptivesam,
      title={AdaptiveSAM: Towards Efficient Tuning of SAM for Surgical Scene Segmentation}, 
      author={Jay N. Paranjape and Nithin Gopalakrishnan Nair and Shameema Sikder and S. Swaroop Vedula and Vishal M. Patel},
      year={2023},
      eprint={2308.03726},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}