HORNet: Task-Guided Frame Selection for Video Question Answering with Vision-Language Models

• URL: http://arxiv.org/abs/2603.18850v1
• Date: Thu, 19 Mar 2026 12:53:32 GMT
• Title: HORNet: Task-Guided Frame Selection for Video Question Answering with Vision-Language Models
• Abstract summary: HorNet reduces input frames by up to 99% and VLM processing time by up to 93%.<n>We formalize this as Select Any Frames (SAF), a task that decouples visual input from VLM reasoning.<n>HorNet's policy further transfers across VLM answerers without retraining, yielding an additional 8.5% relative gain when paired with a stronger model.
• Score: 6.358055069130984
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Video question answering (VQA) with vision-language models (VLMs) depends critically on which frames are selected from the input video, yet most systems rely on uniform or heuristic sampling that cannot be optimized for downstream answering quality. We introduce \textbf{HORNet}, a lightweight frame selection policy trained with Group Relative Policy Optimization (GRPO) to learn which frames a frozen VLM needs to answer questions correctly. With fewer than 1M trainable parameters, HORNet reduces input frames by up to 99\% and VLM processing time by up to 93\%, while improving answer quality on short-form benchmarks (+1.7\% F1 on MSVD-QA) and achieving strong performance on temporal reasoning tasks (+7.3 points over uniform sampling on NExT-QA). We formalize this as Select Any Frames (SAF), a task that decouples visual input curation from VLM reasoning, and show that GRPO-trained selection generalizes better out-of-distribution than supervised and PPO alternatives. HORNet's policy further transfers across VLM answerers without retraining, yielding an additional 8.5\% relative gain when paired with a stronger model. Evaluated across six benchmarks spanning 341,877 QA pairs and 114.2 hours of video, our results demonstrate that optimizing \emph{what} a VLM sees is a practical and complementary alternative to optimizing what it generates while improving efficiency. Code is available at https://github.com/ostadabbas/HORNet.

Related papers

• VisualRouter: Query-Grounded Visual Sampling for Long Video Understanding [60.161767365040554]
  We propose a training-free and plug-and-play framework for query-grounded visual sampling.<n>Visual first classifies each query as either global or local and then applies the corresponding sampling strategy.<n>Experiments show that Visual consistently improves multiple LVLMs over uniform sampling, achieving gains of 5.2%, 7.7%, and 11.6% on Video-MME, LongVideoBench, and MLVU with Qwen2.5-VL-7B, and outperforming existing training-free visual sampling methods under the same setting.
  arXiv  Detail & Related papers  (2026-07-30T16:23:10Z)
• Efficient Frame Selection for Long Videos at Test Time with Attention-Based MLLM Selectors [69.75895255395702]
  Cross-modal attention at validation-selected extraction layers in MLLMs already provides query-relevant frame evidence without requiring autoregressive generation.<n>A lightweight MLLM selector, even with only 2B parameters, can extract frame-level evidence by converting selected-layer attention into relevance scores through query-conditioned aggregation.<n>Under a 32-frame budget, our selector improves over uniform sampling by up to 6.4 points on Video-MME and outperforms prior training-based selectors under matched frame budgets.
  arXiv  Detail & Related papers  (2026-07-17T07:04:31Z)
• Event-Anchored Frame Selection for Effective Long-Video Understanding [67.56884568828508]
  Event-Anchored Frame Selection (EFS) is a hierarchical, event-aware pipeline.<n>As a training-free, plug-and-play module, EFS can be seamlessly integrated into off-the-shelf LVLMs.
  arXiv  Detail & Related papers  (2026-03-01T08:25:37Z)
• VideoBrain: Learning Adaptive Frame Sampling for Long Video Understanding [9.415923244280542]
  VideoBrain is an end-to-end framework that enables Vision-Language Models to adaptively acquire visual information through learned sampling policies.<n>Our approach features dual complementary agents: a CLIP-based agent for semantic retrieval across the video and a Uniform agent for dense temporal sampling within intervals.
  arXiv  Detail & Related papers  (2026-02-04T00:08:35Z)
• Temporal Chain of Thought: Long-Video Understanding by Thinking in Frames [70.93346841539626]
  We present Temporal Chain of Thought, an inference strategy for video question-answering.<n>We use the VLM itself to iteratively identify and extract the most relevant frames from the video.<n>We demonstrate how leveraging more computation at inference-time to select the most relevant context leads to improvements in accuracy.
  arXiv  Detail & Related papers  (2025-07-01T18:39:26Z)
• BOLT: Boost Large Vision-Language Model Without Training for Long-form Video Understanding [51.49345400300556]
  Large video-language models (VLMs) have demonstrated promising progress in various video understanding tasks.<n>Traditional approaches, such as uniform frame sampling, often inevitably allocate resources to irrelevant content.<n>We introduce BOLT, a method to BOost Large VLMs without additional Training through a comprehensive study of frame selection strategies.
  arXiv  Detail & Related papers  (2025-03-27T13:18:40Z)
• Too Many Frames, Not All Useful: Efficient Strategies for Long-Form Video QA [40.21221568678641]
  Long-form videos that span across wide temporal intervals are highly information redundant.<n>All information necessary to generate a correct response can often be contained within a small subset of frames.<n>Recent literature explore use of large language models in LVQA benchmarks, achieving exceptional performance.
  arXiv  Detail & Related papers  (2024-06-13T17:59:16Z)
• Plug-and-Play VQA: Zero-shot VQA by Conjoining Large Pretrained Models with Zero Training [82.30343537942608]
  We propose Plug-and-Play VQA (PNP-VQA), a modular framework for zero-shot VQA. We first generate question-guided informative image captions, and pass the captions to a PLM as context for question answering. PNP-VQA achieves state-of-the-art results on zero-shot VQAv2 and GQA.
  arXiv  Detail & Related papers  (2022-10-17T06:29:54Z)
• FAST-VQA: Efficient End-to-end Video Quality Assessment with Fragment Sampling [54.31355080688127]
  Current deep video quality assessment (VQA) methods are usually with high computational costs when evaluating high-resolution videos. We propose Grid Mini-patch Sampling (GMS), which allows consideration of local quality by sampling patches at their raw resolution. We build the Fragment Attention Network (FANet) specially designed to accommodate fragments as inputs. FAST-VQA improves state-of-the-art accuracy by around 10% while reducing 99.5% FLOPs on 1080P high-resolution videos.
  arXiv  Detail & Related papers  (2022-07-06T11:11:43Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.