metadata
pipeline_tag: robotics
tags:
- robotics
- human-to-robot
- video-generation
- inverse-dynamics
RoboTryOn
Checkpoints for Synthesizing Dexterous Robot Demonstrations from Human Videos via Virtual Try-On.
Paper · Project page · Code
| Folder | Model | Base |
|---|---|---|
image/ |
Keyframe translation (LoRA) | FLUX.2-klein-base-9B |
video/ |
Video synthesis (in-context LoRA) | LTX-2.5-Diffusers |
idm/ |
Inverse dynamics model, shared by all embodiments | Depth-Anything-V2-Large |
All models are trained on teleoperated episodes of OpenArm, RB-Y1 and GR-1.
Usage
hf download RyanL22/RoboTryOn --local-dir ckpt
python scripts/run_pipeline.py --config configs/pipeline.yaml ... \
checkpoints.image=ckpt/image checkpoints.video=ckpt/video checkpoints.idm=ckpt/idm
Citation
@article{lee2026robotryon,
title={Synthesizing Dexterous Robot Demonstrations from Human Videos via Virtual Try-On},
author={Lee, Seungyong and Jang, Suhyeok and Kim, Minyoung and Shin, Jinwoo},
journal={arXiv preprint arXiv:2611.05163},
year={2026}
}