An annotation tool for creating labeled segmentation datasets.
Clone the repository locally:
git clone https://github.com/Steven-Holland/segment-annotator.git
Create the conda env using python>=3.10
. Please follow the instructions here to install both PyTorch and TorchVision dependencies. Installing both PyTorch and TorchVision with CUDA support is strongly recommended.
conda create -n Annotator python=3.10
conda activate Annotator
Install the packages:
cd segment-annotator
pip install -r requirements.txt
Download SAM and FastSAM model checkpoints. Once they are downloaded, move them to ../src/assets/models
.
Move into the src
folder and run main.py
cd src
python main.py