IamSAM

A small project for comparing the outputs of SAM (Segment Anything [1]) and FastSAM [2], with a practical GUI (Pyside6)

The project is still in pre-release, so do not hesitate to send your recommendations or the bugs you encountered!

Principle

The user can upload any picture, then add 'positive' and 'negative' points that will be used as prompt in FASTSAM or SAM algorithm. The tool creates a small output mask (jpg) in the folder, and allows the user to directly visualize the segmentation results.

Installation instructions

1. Clone the repository:

git clone https://github.com/s-du/IamSAM

2. Navigate to the app directory:

cd IamSAM

3. (Optional) Install and activate a virtual environment

4. Install the required dependencies: Please note that SEGMENT ANYTHING requires a CUDA environment (the user must manually install torch with the appropriate CUDA version). Models checkpoints should be downloaded manually and placed in resources/other:

• FastSAM-x.pt
• sam_vit_h_4b8939.pth

6. Run the app:

python main.py

References

[1] See https://github.com/facebookresearch/segment-anything

[2] See https://github.com/CASIA-IVA-Lab/FastSAM

@misc{zhao2023fast,
title={Fast Segment Anything},
author={Xu Zhao and Wenchao Ding and Yongqi An and Yinglong Du and Tao Yu and Min Li and Ming Tang and Jinqiao Wang},
year={2023},
eprint={2306.12156},
archivePrefix={arXiv},
primaryClass={cs.CV}
}