datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
EgoTextVQA-Indoor
EgoTextVQA-Indoor
Indoor split of EgoTextVQA,
an egocentric scene-text VQA benchmark (Zhang et al., 2025). 2,216 open-ended
questions over first-person indoor video clips that require reading text in
the scene to answer.
Format
test.json is a list of records:
{
"video_id": "0479bea8-d221-4c6a-8c91-60108e43fe56",
"question_type": "book-related",
"correct_answer": "the book is being placed on the shelf",
"question": "where is the book titled fables?"
}
Videos… See the full description on the dataset page: https://huggingface.co/datasets/nv-njb/EgoTextVQA-Indoor.han-simple-indoor-navigation-v1
Simple Indoor Navigation Dataset
Basic directional commands
for indoor humanoid navigation training.
License
MIT
indoor-navigation-json
Indoor Navigation Dataset (JSON)
This dataset maps human navigation instructions to indoor room targets.
It is designed for humanoid robot navigation training.