EgoThink: Evaluating First-Person Perspective Thinking Capability of Vision-Language Models

1 code implementation • 27 Nov 2023 • Sijie Cheng, Zhicheng Guo, Jingwen Wu, Kechen Fang, Peng Li, Huaping Liu, Yang Liu

However, the capability of VLMs to "think" from a first-person perspective, a crucial attribute for advancing autonomous agents and robotics, remains largely unexplored.

Attribute Question Answering +1