GRAID BDD100K Question-Answer Dataset

GRAID BDD100K Question-Answer Dataset Paper Project Page Overview This dataset was generated using GRAID (Generating Reasoning questions from Analysis of Images via Discriminative artificial intelligence), a framework for creating spatial reasoning datasets from object detection annotations. GRAID transforms raw object detection data into structured question-answer pairs that test various aspects of object localization, visual reasoning, spatial reasoning, and object… See the full description on the dataset page:

Source: Hugging Face Hub (kd7/graid-bdd). Metadata imported from the dataset’s Hub tags.