Closed healthonrails closed 1 year ago
The high-quality object masks generated by the Segment Anything Model (SAM) can be produced from input prompts like points or boxes, and can cover all objects in an image. To access the SAM and utilize its capabilities, you can visit the following GitHub repository: https://github.com/facebookresearch/segment-anything. You can access the web demo site for the Segment Anything Model at https://segment-anything.com/.
The code requires Python version 3.8 or higher, along with PyTorch version 1.7 or higher, as well as TorchVision version 0.8 or higher. To install these dependencies, kindly refer to the instructions provided here. It's strongly recommended to install both PyTorch and TorchVision with CUDA support. To install Segment Anything, run the following command in your terminal: pip install git+https://github.com/facebookresearch/segment-anything.git
Done
One example case is from here https://github.com/originlake/labelme-with-segment-anything.