Wassump homie it's Tony.
Once you clone this repo, there's a bit more you gotta do:
Download the model here: https://dl.fbaipublicfiles.com/segment_anything/sam_vit_b_01ec64.pth
This should be put into the root.
Make a venv. Then run pip install -r requirements.txt.
In the venv, run this first, to make sure you're using CUDA
python a.py
If it says false, run dis.
pip install --force-reinstall torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
Try to run train.py ig:
python train.py