ECCV 2024
- Aleksandar (Suny) Shtedritski
- Christian Rupprecht
- Andrea Vedaldi
Contact: {suny, chrisr, vedaldi}@robots.ox.ac.uk
For more detailed information about SHIC, visit our project website.
A demo of SHIC is available on Hugging Face. You can try it out here.
conda create -n shicenv python=3.9
conda activate shicenv
pip install torch==1.12.0 torchvision==0.13.0 einops spharapy==1.1.2 trimesh==3.23.5 matplotlib numpy scipy Pillow jupyter
Download pretrained the weights here here.
Unzip and put in this folder. Should look like
.
├── ...
├── models
│ ├── shapes # obj shape files
│ └── weights # pth weights for shape features and image models
├── README.md
└── ...
Run the shic.ipynb
notekbook for a more custom demo of the model. The notebook runs very quickly on a CPU, which is used by default
Coming soon!
@inproceedings{shtedritski2024SHIC,
title={SHIC: Shape-Image Correspondences with no Keypoint Supervision},
author={Shtedritski, Aleksandar and Rupprecht, Christian and Vedaldi, Andrea},
year={2024},
booktitle={ECCV},
}