Do You See What I Am Pointing At? Gesture-Based Egocentric Video Question Answering

arXiv cs.CV / 3/16/2026

📰 NewsTools & Practical UsageModels & Research

共有:

Key Points

Introduces EgoPointVQA, a dataset and benchmark for gesture-grounded egocentric question answering, comprising 4000 synthetic and 400 real-world videos across multiple deictic reasoning tasks.
Proposes Hand Intent Tokens (HINT) derived from 3D hand keypoints and interleaves them with model input to provide explicit spatial and temporal context for interpreting pointing intent.
Demonstrates that HINT improves performance across backbones and model sizes, with HINT-14B achieving 68.1% accuracy on average over 6 tasks, surpassing the state-of-the-art InternVL3-14B by 6.6%.
Will release the code, model, and dataset to open research, with a project page at https://yuuraa.github.io/papers/choi2026egovqa.
Addresses gaps in gesture-rich data for egocentric AI assistants and advances gesture-based VQA by enabling more accurate understanding of pointing gestures.

Abstract

Understanding and answering questions based on a user's pointing gesture is essential for next-generation egocentric AI assistants. However, current Multimodal Large Language Models (MLLMs) struggle with such tasks due to the lack of gesture-rich data and their limited ability to infer fine-grained pointing intent from egocentric video. To address this, we introduce EgoPointVQA, a dataset and benchmark for gesture-grounded egocentric question answering, comprising 4000 synthetic and 400 real-world videos across multiple deictic reasoning tasks. Built upon it, we further propose Hand Intent Tokens (HINT), which encodes tokens derived from 3D hand keypoints using an off-the-shelf reconstruction model and interleaves them with the model input to provide explicit spatial and temporal context for interpreting pointing intent. We show that our model outperforms others in different backbones and model sizes. In particular, HINT-14B achieves 68.1% accuracy, on average over 6 tasks, surpassing the state-of-the-art, InternVL3-14B, by 6.6%. To further facilitate the open research, we will release the code, model, and dataset. Project page: https://yuuraa.github.io/papers/choi2026egovqa