AutoSAM

AutoSAM: Adapting SAM to Medical Images by Overloading the Prompt Encoder

Overview

This work improves the Segment Anything Model (SAM) for medical image segmentation by replacing its conditioning mechanism with an image-based encoder. Without further fine-tuning SAM, this modification achieves state-of-the-art results on medical images and video benchmarks.

Paper

The paper associated with this repository can be found here.

Datasets

We used the following datasets in our experiments:

monu glas polyp

SAM checkopints

sam base sam large sam huge

Usage

To use AutoSAM, follow these steps:

Clone the repository:

git clone https://github.com/your_username/AutoSAM.git
cd AutoSAM/

conda:

conda create --name autosam python=3.10
pip install -r requirements.txt

training:
```
python train.py
```