Saved in:
Bibliographic Details
Main Authors: Xu, Yinsong, Jing, Wei, Zhang, Liuxin, Lv, Wanjun, Li, Hui
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.29402
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910269043638272
author Xu, Yinsong
Jing, Wei
Zhang, Liuxin
Lv, Wanjun
Li, Hui
author_facet Xu, Yinsong
Jing, Wei
Zhang, Liuxin
Lv, Wanjun
Li, Hui
contents Understanding long-form egocentric videos remains challenging for multimodal large language models (MLLMs) due to limited context length and insufficient grounding of fine-grained visual details. The recently proposed HD-EPIC benchmark highlights these limitations: even strong long-context models achieve relatively low performance across diverse video question answering tasks. In this paper, we propose a unified framework that decouples long-video reasoning into two complementary forms of evidence: semantic evidence and visual evidence. Semantic evidence captures global procedural structure through a coarse-to-fine extraction pipeline, while object-centric visual evidence preserves fine-grained grounding through bounding boxes and visual embeddings. During inference, we formulate reasoning as a query-conditioned evidence retrieval and integration process, dynamically selecting relevant information from both sources. Our approach achieves competitive performance in the HD-EPIC-VQA Challenge across multiple task categories. More broadly, our results demonstrate that explicitly structuring, retrieving, and integrating semantic and visual evidence is critical for effective long-video understanding with MLLMs.
format Preprint
id arxiv_https___arxiv_org_abs_2605_29402
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Semantic and Visual Evidence for Efficient Long-Video Reasoning: A Solution for the HD-EPIC VQA Challenge
Xu, Yinsong
Jing, Wei
Zhang, Liuxin
Lv, Wanjun
Li, Hui
Computer Vision and Pattern Recognition
Artificial Intelligence
Understanding long-form egocentric videos remains challenging for multimodal large language models (MLLMs) due to limited context length and insufficient grounding of fine-grained visual details. The recently proposed HD-EPIC benchmark highlights these limitations: even strong long-context models achieve relatively low performance across diverse video question answering tasks. In this paper, we propose a unified framework that decouples long-video reasoning into two complementary forms of evidence: semantic evidence and visual evidence. Semantic evidence captures global procedural structure through a coarse-to-fine extraction pipeline, while object-centric visual evidence preserves fine-grained grounding through bounding boxes and visual embeddings. During inference, we formulate reasoning as a query-conditioned evidence retrieval and integration process, dynamically selecting relevant information from both sources. Our approach achieves competitive performance in the HD-EPIC-VQA Challenge across multiple task categories. More broadly, our results demonstrate that explicitly structuring, retrieving, and integrating semantic and visual evidence is critical for effective long-video understanding with MLLMs.
title Semantic and Visual Evidence for Efficient Long-Video Reasoning: A Solution for the HD-EPIC VQA Challenge
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2605.29402