dabblle/segment-anything-webui

Yet another SAM web UI

Segment Anything web UI

This is a web interface for the Segment Anything.

Usage

Environment Require: Python >= 3.8.13, Node >= 18.15.0 (LTS), CUDA(optional)

1. Fowllow the instructions in the Segment Anything to install

# e.g.
pip install git+https://github.com/facebookresearch/segment-anything.git
pip install opencv-python pycocotools matplotlib onnxruntime onnx

mkdir model
# download the model to `model/`
wget https://dl.fbaipublicfiles.com/segment_anything/sam_vit_b_01ec64.pth -O model/sam_vit_b_01ec64.pth
# https://dl.fbaipublicfiles.com/segment_anything/sam_vit_l_0b3195.pth
# https://dl.fbaipublicfiles.com/segment_anything/sam_vit_h_4b8939.pth

2. Install the webui dependencies:

# python server as backend
pip3 install torch numpy 'uvicorn[standard]' fastapi pydantic python-multipart Pillow click
# or 
cd script && pip3 install -r requirements.txt
# webui frontend
npm i

3. run the server:

python3 script/server.py

4. run the webui:

npm run dev

Advanced

Change the .env.local file to change the server address.

The model server can be run on a remote GUI server, and the webui can be run on a local machine.

The API in server.py is lambda function. Though it is slow (Encoding Image Each Request), it is easy to deploy.

Upload Image on

TODO

• Pre extract image features
• Frontend onnx inference
• Add CLIP for text Prompt
• Better compress for mask matrix

Reference

• Segment Anything | Meta AI
• facebookresearch/segment-anything: The repository provides code for running inference with the SegmentAnything Model (SAM), links for downloading the trained model checkpoints, and example notebooks that show how to use the model.

License

MIT