HOI-R1 Rex-Omni

This is the HOI-R1 model fine-tuned from IDEA-Research/Rex-Omni for human-object interaction detection on HICO-DET. It contains the inference-ready weights from training checkpoint 1176, together with its model configuration, tokenizer, image preprocessor, and chat template. The DeepSpeed optimizer and training state are not included.

The Rex-Omni training data is in the Rex-Omni/ folder of HOI-R1 HICO-DET dataset.

The base model and this derivative are subject to the IDEA License 1.0, which permits non-commercial research use and has redistribution conditions. The license text is included in this repository.

Citation

@article{chen2025hoi,
  title={HOI-R1: Exploring the Potential of Multimodal Large Language Models for Human-Object Interaction Detection},
  author={Chen, Junwen and Xiong, Peilin and Yanai, Keiji},
  journal={arXiv preprint arXiv:2510.05609},
  year={2025}
}