Reproduce “Benchmarking Multimodal Large Language Models for Zero-Shot Human-Object Interaction” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Oriol Rabasseda, Zenjie Li, Kamal Nasrollahi, Sergio Escalera

The authors evaluate multimodal language models for human–object interaction recognition and detection without HOI-specific fine-tuning. An external object detector filters object classes or proposes pairs, while yes/no prompts and visual context let the model judge interactions; token probabilities become scores. On HICO, Qwen2-VL 2B reaches 53.50 mAP with Objects as Context, versus 51.39 baseline, within fixed categories and detector limits.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.