Chameleon: A Data-Efficient Generalist for Dense Visual Prediction in the Wild

This repository contains official code for Chameleon: A Data-Efficient Generalist for Dense Visual Prediction in the Wild (ECCV 2024 oral).

The documentation for downloading and preprocessing datasets will be uploaded soon.

Please checkout gaudi-v2 branch for gaudi-v2 implementation of Chameleon.

Citation

If you find this work useful, please consider citing:

@article{kim2024chameleon,
  title={Chameleon: A Data-Efficient Generalist for Dense Visual Prediction in the Wild},
  author={Kim, Donggyun and Cho, Seongwoong and Kim, Semin and Luo, Chong and Hong, Seunghoon},
  journal={arXiv preprint arXiv:2404.18459},
  year={2024}
}