This dataset is presented in the paper GRAID: Enhancing Spatial Reasoning of VLMs Through High-Fidelity Data Generation.
Project page:
https://ke7.github.io/graid/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… See the full description on the dataset page:
https://huggingface.co/datasets/kd7/graid-nuimages-wd.
