zooblastlbz/Bench-SPACE

Bench-SPACE Bench-SPACE is the dataset of the SPACE benchmark from "Human-Centric Image Captioning with Subject-Centered Spatial Understanding". It evaluates how well vision-language models describe human subjects in images, focusing on Spatial & Orientation, Action & Pose, Appearance, and Multi-person Interaction. Model captions are scored by an LLM judge against per-image QA pairs derived from human-verified descriptions. 2,981 images (~4.3 GB) 19,892 QA pairs across 16… See the full description on the dataset page: https://huggingface.co/datasets/zooblastlbz/Bench-SPACE.