Skip to content

Latest commit

 

History

History
29 lines (20 loc) · 1.33 KB

README.md

File metadata and controls

29 lines (20 loc) · 1.33 KB

ZeroShotSAM

Overview

ZeroShotSAM is a project focused on zero-shot medical image segmentation using a sparse prompt approach, leveraging a finetuned version of the Segment Anything Model (SAM). This repository contains code and resources for finetune and evaluating the model, as well as documentation to help users get started.

Features

  • Utilizes the Segment Anything Model (SAM) for medical image segmentation.
  • Implements zero-shot segmentation using sparse prompts.
  • Provides a finetuning mechanism to adapt SAM to medical domain data.
  • Includes a script for finetuning and evaluation.

Datasets

We used the following datasets in our experiments:

monu glas

SAM checkopints

sam base sam large

Usage

CUDA_VISIBLE_DEVICES=0 python -W ignore train.py --task glas --vit vit_b --epoches 100
CUDA_VISIBLE_DEVICES=1 python -W ignore train.py --task glas --vit vit_l --epoches 100
CUDA_VISIBLE_DEVICES=2 python -W ignore train.py --task monu --vit vit_b --epoches 100