E2083 - Show, Don't Tell: Evaluating Spatial Cognition in Generative Pixels Rather Than LLM Text

🤗 Upvotes: 34 | cs.CV

Authors:
Xu Wang, Kaixiang Yao, Miao Pan, Xiaohe Zhou, Xuanyu Liu, Wenqi Zhang, Xuhong Zhang

Title:
Show, Don't Tell: Evaluating Spatial Cognition in Generative Pixels Rather Than LLM Text

Arxiv:
http://arxiv.org/abs/2607.21072v1

Abstract:
Spatial intelligence is essential for agents to move from static semantic understanding toward interacting with the physical world. Many spatial tasks are grounded in continuous visual scenes, where locations, regions, and paths are more naturally expressed by pointing, marking, or drawing than by reporting precise coordinates or discrete tex...