This is the code repo for ICPR2024 paper Sketch2Seg: Sketch-based Image Segmentation with Pre-trained Diffusion Model
To install the required python packages to run Sketch2Seg with conda
conda env create -f environment.yaml
conda activate control
To evaluation sketch2seg, please prepare validation dataset with photo,sketch and seg annotation,and replace in the evaluation.py.
To run a segmentation, you can run
python evaluation.py --exp_dir YOUR_CHECKPOINT_PATH --save_dir YOUR_SAVE_PATH
If you want to train your own model, please prepare your training data as well, then you can run
python train_classifier.py --exp_dir PATH_TO_SAVE_CHECKPOINT
You can get the dataset at here We provide the 256*256 resolution version,if you need other resolutions,please contact us([email protected])
Thanks for the wanderful work ControlNet controlnet. We also want to thank Baranchuk's work LABEL-EFFICIENT SEMANTIC SEGMENTATION Thanks Conghui's work Sketch-a-Segmenter