Bridging VideoQA and Video-Guided Agentic Tasks via Generalized Keyframe Extraction
Summary
This paper introduces VG-GUIBench, a benchmark to evaluate MLLM-based GUI agents' ability to follow video tutorials, and proposes TASKER, a keyframe extraction method that improves performance on VideoQA and video-guided agentic tasks.
View Cached Full Text
Cached at: 06/30/26, 03:33 AM
Paper page - Bridging VideoQA and Video-Guided Agentic Tasks via Generalized Keyframe Extraction
Source: https://huggingface.co/papers/2606.29445
Abstract
A new benchmark evaluates multimodal large language models’ ability to understand video content and perform GUI tasks, while a novel keyframe extraction method improves performance on both video question answering and video-guided agentic tasks.
Video understanding is a fundamental capability for multimodal intelligence, and recentMultimodal Large Language Models(MLLMs) have achieved remarkable performance onVideo Question Answering(VideoQA) benchmarks. However, existing benchmarks primarily evaluate whether models can perceive shallow visual cues, while rarely examining whether MLLMs can learn deeper knowledge or procedural skills fromvideo tutorialsand generalize them to downstream long-horizon agentic tasks. To address this gap, we introduce VG-GUIBench (Video-Guided GUI Benchmark), a new benchmark designed to evaluate whether MLLM-basedGUI agentscan followvideo tutorialsto complete corresponding GUI interactive tasks. Furthermore, we observe that the performance of models on both VideoQA and video-guided agentic tasks critically depends on effectivekeyframe extraction. Based on this observation, we propose TASKER (Task-driven And Scene-aware Keyframe searchER), akeyframe extractionalgorithm that jointly considerstask relevanceandscene dynamicsto identify informative frames. Experimental results demonstrate that TASKER achieves significant performance improvements on both VideoQA and video-guided agentic task benchmarks, outperforming the best baseline by 2.0% on the EgoSchema fullset and 1.8% on the NExT-QA dataset, respectively. These results further highlight the potential of generalizedkeyframe extractionmethods for video understanding tasks. Our code and data are available at https://github.com/VG-GUI-TASKER/VG-GUI-TASKER.
View arXiv pageView PDFProject pageGitHub5Add to collection
Get this paper in your agent:
hf papers read 2606\.29445
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2606.29445 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2606.29445 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2606.29445 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
Video2GUI: Synthesizing Large-Scale Interaction Trajectories for Generalized GUI Agent Pretraining
Proposes Video2GUI, a framework to automatically extract GUI interaction trajectories from unlabeled instructional videos, building WildGUI dataset with 12M trajectories across 1500+ apps. Pre-training on this data yields 5-20% improvements on GUI grounding and action benchmarks.
GUICrafter: Weakly-Supervised GUI Agent Leveraging Massive Unannotated Screenshots
GUICrafter introduces a weakly-supervised GUI agent that leverages massive unannotated screenshots and a two-stage curriculum learning framework to reduce reliance on expensive human annotations, achieving competitive performance with advanced systems like UI-TARS using only 0.1% of its data.
Teach-and-Repeat: Accurately Extracting Operational Knowledge from Mobile Screen Demonstrations to Empower GUI Agents
Introduces Teach VLM, a model that extracts step-by-step operational knowledge from mobile screen demonstrations, and the Teach-and-Repeat paradigm that uses this knowledge to guide GUI agents, achieving state-of-the-art performance on a new benchmark.
VGenST-Bench: A Benchmark for Spatio-Temporal Reasoning via Active Video Synthesis
VGenST-Bench is a benchmark that uses generative models to actively synthesize controlled spatio-temporal reasoning scenarios, with a multi-agent pipeline and human quality control, to evaluate multimodal large language models.
VEFX-Bench: A Holistic Benchmark for Generic Video Editing and Visual Effects
VEFX-Bench introduces a large-scale human-annotated video editing dataset (5,049 examples) with multi-dimensional quality labels and a specialized reward model for standardized evaluation of video editing systems. The paper addresses the lack of comprehensive benchmarks in AI-assisted video creation by providing VEFX-Dataset, VEFX-Reward, and a 300-video-prompt benchmark that reveals gaps in current editing models.