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

TL;DR

EgoThink introduces a new benchmark for evaluating vision-language models' ability to understand and answer questions from a first-person perspective using egocentric video clips, highlighting current limitations and potential improvements.

Contribution

The paper presents EgoThink, a novel first-person perspective visual question-answering benchmark, and evaluates 18 VLMs, revealing significant room for enhancement in first-person understanding.

Findings

GPT-4V outperforms other models in many dimensions.

Enlarging trainable parameters improves model performance.

All models show potential for growth in first-person tasks.

Abstract

Vision-language models (VLMs) have recently shown promising results in traditional downstream tasks. Evaluation studies have emerged to assess their abilities, with the majority focusing on the third-person perspective, and only a few addressing specific tasks from the first-person perspective. However, the capability of VLMs to "think" from a first-person perspective, a crucial attribute for advancing autonomous agents and robotics, remains largely unexplored. To bridge this research gap, we introduce EgoThink, a novel visual question-answering benchmark that encompasses six core capabilities with twelve detailed dimensions. The benchmark is constructed using selected clips from egocentric videos, with manually annotated question-answer pairs containing first-person information. To comprehensively assess VLMs, we evaluate eighteen popular VLMs on EgoThink. Moreover, given the open-ended format of the answers, we use GPT-4 as the automatic judge to compute single-answer grading. Experimental results indicate that although GPT-4V leads in numerous dimensions, all evaluated VLMs still possess considerable potential for improvement in first-person perspective tasks. Meanwhile, enlarging the number of trainable parameters has the most significant impact on model performance on EgoThink. In conclusion, EgoThink serves as a valuable addition to existing evaluation benchmarks for VLMs, providing an indispensable resource for future research in the realm of embodied artificial intelligence and robotics.