naver-iv/MicroMat-3K

ZIM: Zero-Shot Image Matting for Anything

Introduction

🚀 Introducing ZIM: Zero-Shot Image Matting – A Step Beyond SAM! 🚀

While SAM (Segment Anything Model) has redefined zero-shot segmentation with broad applications across multiple fields, it often falls short in delivering high-precision, fine-grained masks. That’s where ZIM comes in.

🌟 What is ZIM? 🌟

ZIM (Zero-Shot Image Matting) is a groundbreaking model developed to set a new standard in precision matting while maintaining strong zero-shot capabilities. Like SAM, ZIM can generalize across diverse datasets and objects in a zero-shot paradigm. But ZIM goes beyond, delivering highly accurate, fine-grained masks that capture intricate details.

🔍 Get Started with ZIM 🔍

Ready to elevate your AI projects with unmatched matting quality? Access ZIM on our project page, Arxiv, and Github.

Installation

pip install zim_anything

or

git clone https://github.com/naver-ai/ZIM.git
cd ZIM; pip install -e .

Usage

1. Make the directory zim_vit_l_2092.
2. Download the encoder weight and decoder weight.
3. Put them under the zim_vit_b_2092 directory.

from zim_anything import zim_model_registry, ZimPredictor

backbone = "vit_l"
ckpt_p = "zim_vit_l_2092"

model = zim_model_registry[backbone](checkpoint=ckpt_p)
if torch.cuda.is_available():
    model.cuda()

predictor = ZimPredictor(model)
predictor.set_image(<image>)
masks, _, _ = predictor.predict(<input_prompts>)

Citation

If you find this project useful, please consider citing:

@article{kim2024zim,
  title={ZIM: Zero-Shot Image Matting for Anything},
  author={Kim, Beomyoung and Shin, Chanyong and Jeong, Joonhyun and Jung, Hyungsik and Lee, Se-Yun and Chun, Sewhan and Hwang, Dong-Hyun and Yu, Joonsang},
  journal={arXiv preprint arXiv:2411.00626},
  year={2024}
}