Self-supervised learning framework for establishing dense correspondences between photographs and sketches, enabling automatic image warping onto sketch templates. Built with PyTorch.
- Implements the paper
Learning Dense Correspondences between Photos and Sketches - Self-supervised training and evaluation pipeline
- Image-to-Sketch warping capabilities
- PyTorch-based implementation
Create the conda environment:
conda env create -f environment.ymlActivate the environment:
conda activate warp_envOr instead of the above, you can simply:
bash scripts/setup.shEnvironment requirements are specified in environment.yml. Key dependencies include:
- PyTorch >= 2.5
- Python >= 3.10
- CUDA (optional for inference, for GPU support)
This repository is licensed under the MIT License, allowing for its use in research and educational purposes.