A Multimodal Dataset for Vision-Language Grounding in Human-Robot Interaction (HRI)📍 Language: Japanese | 🤖 Focus: Embodied AI | 📆 Size: 142 Scenes | 🔬 Granularity: Object-Level
J-ORA (Japanese Object Reference and Action) is a multimodal benchmark for grounded vision-language learning in robotics. It is designed for understanding Japanese… See the full description on the dataset page:
https://huggingface.co/datasets/atamiles/J-ORA.
