ViT-Up

ViT-Up is an implicit feature upsampler for Vision Transformers that predicts backbone-aligned features at arbitrary continuous image coordinates.

This repository provides pretrained ViT-Up weights for DINOv3-S+ and DINOv3-B.

• Paper: https://arxiv.org/abs/2606.14024
• HF Paper page: https://huggingface.co/papers/2606.14024
• Project page: https://vitup.papers.discuna.com/
• Code: https://github.com/krispinwandel/vit-up

Citation

@misc{wandel2026vitupfaithfulfeatureupsampling,
      title={ViT-Up: Faithful Feature Upsampling for Vision Transformers}, 
      author={Krispin Wandel and Jingchuan Wang and Hesheng Wang},
      year={2026},
      eprint={2606.14024},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2606.14024}, 
}