EgoThink: Evaluating First-Person Perspective Thinking Capability of Vision-Language Models
This provides a new benchmark for evaluating first-person perspective capabilities in vision-language models, which is incremental but important for advancing embodied AI and robotics.
The authors tackled the lack of evaluation for vision-language models' first-person perspective thinking, a key capability for autonomous agents, by introducing EgoThink, a benchmark based on egocentric videos with manually annotated questions, and found that while GPT-4V performed best, all models have significant room for improvement, with model size being the most impactful factor.
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.