PixelSAM is a is a graphical image annotation tool powered by the Segment Anything Model

The tool allows users to segment the images automatically making the labelling process more smooth.

It is based on Python and uses Tkinter for the UI.

Install Segment Anything:

pip install git+https://github.com/facebookresearch/segment-anything.git

Clone the PixelSAM Repository:

git clone [email protected]:Rahul-Pi/PixelSAM.git

Download the Model Checkpoints:

cd PixelSAM
wget https://dl.fbaipublicfiles.com/segment_anything/sam_vit_h_4b8939.pth

Note: If the link does not work or you want to choose a different model go the segment anything repository and download.

To Run:

python3 PixelSAM.py

Click on the load button and select the folder containing the images.

  --model_path: path to model if not in the same folder as the script
    (default='sam_vit_h_4b8939.pth')
  --model_type: type of the model checkpoint used
    (default='vit_h')

Send a pull request

• Save the labels
• Save the segmentation
• Label multiple objects
• Improve the polygon detection
• Possibility to zoom
• Drawing bounding box
• Pre-process batch