theia tiny patch16 224 cddsv EU AI Act Compliance Profile

Theia

The AI Institute

Theia is a vision foundation model for robot learning that distills multiple off-the-shelf vision foundation models trained on varied vision tasks. Theia’s rich visual representations encode diverse visual knowledge, enhancing downstream robot learning. It was introduced in the paper Theia: Distilling Diverse Vision Foundation Models for Robot Learning, which also includes experiments demonstrating that Theia outperforms its teacher models and prior robot learning models using less training data and smaller model sizes. Demo videos can be found on the project page.

Model Details

The theia-tiny-patch16-224-cddsv model, uses DeiT-Tiny as a backbone, and simulatenously distills CLIP, Depth Anything, DINOv2, Segment Anything and ViT. For more information on usage, please visit the Theia repository.

Citation

If you use Theia in your research, please use the following BibTeX entry:

@article{shang2024theia,
  author    = {Shang, Jinghuan and Schmeckpeper, Karl and May, Brandon B. and Minniti, Maria Vittoria and Kelestemur, Tarik and Watkins, David and Herlant, Laura},
  title     = {Theia: Distilling Diverse Vision Foundation Models for Robot Learning},
  journal   = {arXiv},
  year      = {2024},
}

Usage

The pre-trained model weights and code released with Theia are available for use under [The AI Institute License](https://raw.githubusercontent.com/bdaiinstitute/

Your obligations as a theia tiny patch16 224 cddsv deployer

Track your progress

Sign up to track obligations, receive deadline reminders, and generate audit packages.

Tags

Library