Skip to content

Query-aligned video frame selection for long video understanding

Sep 2026 · arXiv.org · Vol abs/2609.31668 · 1 citation · 20 references
Computer Science

TL;DR

This work introduces a frame-selection method designed specifically for multiple-choice questions, and is the first to directly utilize answer choices as inference-time cues for selecting frames relevant to answering a question.

Abstract

Multimodal large language models (MLLMs) process multimodal inputs by converting text, images, and videos into token sequences that are subsequently processed by a backbone language model. While MLLMs have achieved excellent performance in understanding the content of individual images, video understanding remains significantly more difficult because videos contain large number of video frames. MLLMs typically process only a subset of these frames, usually ranging from 8 to 64. MLLMs usually sample frames uniformly, regardless of their relevance to the question being answered. To address this limitation, several training-free, model-agnostic methods for selecting question-relevant frames have recently been proposed. In this work, we introduce a frame-selection method designed specifically for multiple-choice questions. We extend the query text by appending semantic cues derived from the answer choices and employ a direct query-frame alignment scoring mechanism. To the best of our knowledge, our method is the first to directly utilize answer choices as inference-time cues for selecting frames relevant to answering a question. The method first constructs a compact candidate pool by subsampling video frames at a fixed rate. The frames are then scored according to their maximum cosine similarity across all question-answer pairs to identify the most relevant frames for a given query. This approach preserves a fixed token budget while improving the relevance of the visual evidence provided to the downstream MLLM. We evaluate the effectiveness of our frame-selection method on the MLVU, Video-MME, and LongVideoBench benchmarks using three MLLMs: LLaVA-Mini, Qwen2-VL, and LLaVA-Video. Experimental results demonstrate that answer-aware frame selection generally outperforms uniform sampling and existing training-free frame-selection methods under the same frame budget.

View source

Similar papers

#artificial intelligence Preprint Sep 2026

Evaluation of MLLM-Agnostic Plug-and-Play Keyframe Selection Methods for Long Video Understanding

Multimodal large language models (MLLMs) cannot process every frame of a long video because of limitations in visual-token and computational budgets. Three primary approaches have been proposed to enhance their long-video understanding capabilities: (i) Retraining an MLLM on a large video corpus and/or extending its in...

Dilip Sarkar, Md. Safayet Islam, Liang-Yu Liang · 0 citations
Sep 2026

MetaSampling: Making Frame Samplers Efficient for Long-Video Question Answering

Frame selection is an important component of long-video question answering (VQA) with Multimodal Large Language Models (MLLMs). Existing frame-selection methods improve over simple top-$k$ embedding retrieval and uniform sampling, but are typically applied under a fixed global selection budget. We introduce \textbf{Met...

Ashim Dahal, Bikramjit Banerjee · 0 citations
#small language model Preprint Aug 2026

Training-Free Temporal Abstraction for General Video Understanding

STITCH is presented, a training-free method that divides a video into semantically meaningful temporal chunks that are computed once per video and reused across tasks, suggesting that reusable temporal abstraction is a promising direction for general video understanding.

Etienne Casanova, S. Brodjian, Pietro Perona · 0 citations
Open access 2026

VideoEvent: Leveraging Relevance and LLMs for Video Question Answering

VideoEvent is a lightweight and efficient training-free framework that segments an input video into question-relevant temporal events and selectively supplements them with low-level visual cues such as background and object layout, making it highly suitable for practical deployment.

Chen-Chen Lin, Ming-Han Lee, Kun-Ru Wu et al. · 0 citations
Open access 2026

Speaking ITM’s Language: Query Reformulation and Temporal-MMR for Long-Video QA

The resulting training-free pipeline, RECAST, consistently outperforms recent frame-selection baselines without modifying the ITM encoder, and preserves the per-frame matching cost of a standard single-query baseline.

S. Han, Thang Vu, Junyeong Kim · 0 citations

Related blog posts

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.