视频大模型

UniVTG: Towards Unified Video-Language Temporal Grounding

Kevin Qinghong Lin, Pengchuan Zhang, Joya Chen, Shraman Pramanick, Difei Gao, Alex Jinpeng Wang, Rui Yan, Mike Zheng Shou

专题命中 视频理解 ：video-language(title)；分类 cs.CV

Comments Accepted by ICCV 2023. 16 pages, 10 figures, 13 tables. Code: https://github.com/showlab/UniVTG

Towards Long Form Audio-visual Video Understanding

Wenxuan Hou, Guangyao Li, Yapeng Tian, Di Hu

专题命中 视频理解 ：video understanding(title)；分类 cs.MM

Learning Higher-order Object Interactions for Keypoint-based Video Understanding

Yi Huang, Asim Kadav, Farley Lai, Deep Patel, Hans Peter Graf

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Comments SRVU - ICCV' 2021 workshop

Concept Graph Neural Networks for Surgical Video Understanding

Yutong Ban, Jennifer A. Eckhoff, Thomas M. Ward, Daniel A. Hashimoto, Ozanan R. Meireles, Daniela Rus, Guy Rosman

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

SVT: Supertoken Video Transformer for Efficient Video Understanding

Chenbin Pan, Rui Hou, Hanchao Yu, Qifan Wang, Senem Velipasalar, Madian Khabsa

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Procedure-Aware Pretraining for Instructional Video Understanding

Honglu Zhou, Roberto Martín-Martín, Mubbasir Kapadia, Silvio Savarese, Juan Carlos Niebles

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Comments CVPR 2023

Video Swin Transformers for Egocentric Video Understanding @ Ego4D Challenges 2022

Maria Escobar, Laura Daza, Cristina González, Jordi Pont-Tuset, Pablo Arbeláez

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Federated Self-supervised Learning for Video Understanding

Yasar Abbas Ur Rehman, Yan Gao, Jiajun Shen, Pedro Porto Buarque de Gusmao, Nicholas Lane

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Auto-X3D: Ultra-Efficient Video Understanding via Finer-Grained Neural Architecture Search

Yifan Jiang, Xinyu Gong, Junru Wu, Humphrey Shi, Zhicheng Yan, Zhangyang Wang

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Comments Accepted by WACV'2022

Is Space-Time Attention All You Need for Video Understanding?

Gedas Bertasius, Heng Wang, Lorenzo Torresani

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Comments Accepted to ICML 2021

Temporal Query Networks for Fine-grained Video Understanding

Chuhan Zhang, Ankush Gupta, Andrew Zisserman

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Comments Accepted to CVPR 2021(Oral). Project page: http://www.robots.ox.ac.uk/~vgg/research/tqn/

Temporal Aggregate Representations for Long-Range Video Understanding

Fadime Sener, Dipika Singhania, Angela Yao

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Comments ECCV 2020, European Conference on Computer Vision

UniDual: A Unified Model for Image and Video Understanding

Yufei Wang, Du Tran, Lorenzo Torresani

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Long-Term Feature Banks for Detailed Video Understanding

Chao-Yuan Wu, Christoph Feichtenhofer, Haoqi Fan, Kaiming He, Philipp Krähenbühl, Ross Girshick

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Comments Code and models are available at https://github.com/facebookresearch/video-long-term-feature-banks

Efficient Video Understanding via Layered Multi Frame-Rate Analysis

Ziyao Tang, Yongxi Lu, Tara Javidi

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Comments under review

End-to-End Learning of Motion Representation for Video Understanding

Lijie Fan, Wenbing Huang, Chuang Gan, Stefano Ermon, Boqing Gong, Junzhou Huang

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Comments CVPR 2018 spotlight. The first two authors contributed equally to this paper

Large-Scale YouTube-8M Video Understanding with Deep Neural Networks

Manuk Akopyan, Eshsou Khashba

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Comments 6 pages, 5 figures, 3 tables

A generic framework for video understanding applied to group behavior recognition

Sofia Zaidenberg, Bernard Boulay, François Bremond

专题命中 视频理解 ：video understanding(title)；分类 cs.CV

Comments (20/03/2012)

Journal ref 9th IEEE International Conference on Advanced Video and Signal-Based Surveillance (AVSS 2012) (2012) 136 -142

ActAlign: Zero-Shot Fine-Grained Video Classification via Language-Guided Sequence Alignment

Amir Aghdam, Vincent Tao Hu, Björn Ommer

机构 * Department of Computer Science, Temple University（Temple大学计算机科学系） ； Munich Center for Machine Learning（慕尼黑机器学习中心） 纠错

专题命中 视频理解 ：video understanding(abstract)；video-language(abstract)；分类 cs.CV、cs.MM

Comments Accepted to TMLR 2025 - Project page: https://amir-aghdam.github.io/act-align/

Towards Holistic Language-video Representation: the language model-enhanced MSR-Video to Text Dataset

Yuchen Yang, Yingxuan Duan

专题命中 视频理解 ：video understanding(abstract)；video-language(abstract)；分类 cs.CV、cs.MM

Multi-Modal Video Topic Segmentation with Dual-Contrastive Domain Adaptation

Linzi Xing, Quan Tran, Fabian Caba, Franck Dernoncourt, Seunghyun Yoon, Zhaowen Wang, Trung Bui, Giuseppe Carenini

专题命中 视频理解 ：video understanding(abstract)；long video(abstract)；分类 cs.CV、cs.MM

Comments Accepted at the 30th International Conference on Multimedia Modeling (MMM 2024)

From Representation to Reasoning: Towards both Evidence and Commonsense Reasoning for Video Question-Answering

Jiangtong Li, Li Niu, Liqing Zhang

专题命中 视频理解 ：video understanding(abstract)；video reasoning(abstract)；分类 cs.CV、cs.MM

Comments To appear in CVPR 2022

A Video Is Worth 4096 Tokens: Verbalize Videos To Understand Them In Zero Shot

Aanisha Bhattacharya, Yaman K Singla, Balaji Krishnamurthy, Rajiv Ratn Shah, Changyou Chen

专题命中 视频理解 ：video understanding(abstract,comments)；long video(abstract)；分类 cs.CV

Comments Accepted to EMNLP-23 TL;DR: Video understanding lags far behind NLP; LLMs excel in zero-shot. Our approach utilizes LLMs to verbalize videos, creating stories for zero-shot video understanding. This yields state-of-the-art results across five datasets, covering fifteen tasks

DualFact+: A Multimodal Fact Verification Framework for Procedural Video Understanding

DualFact+: 一种用于过程视频理解的多模态事实验证框架

Cennet Oguz, Yasser Hamidullah, Josef van Genabith, Simon Ostermann

机构 * German Research Center for Artificial Intelligence (DFKI)（德国人工智能研究中心（DFKI）） ； Saarland Informatics Campus（萨尔兰信息学校区） 纠错

专题命中 视频理解 ：video understanding(title)

AI总结 DualFact+通过双层多模事实验证框架，针对过程视频描述中的概念事实和上下文事实进行评估，揭示了多模态事实 grounding 的挑战。

Comments ACL 2026 Findings

Hierarchical3D Adapters for Long Video-to-text Summarization

Pinelopi Papalampidi, Mirella Lapata

专题命中 视频理解 ：long video(title)

Wiener Representation Filtering for VLM Hallucination Suppression

用于抑制视觉语言模型幻觉的维纳表示滤波

Ameen Ali, Tamim Zoabi, Lidor Brami, Lior Wolf

机构 * Tel Aviv University（特拉维夫大学） 纠错

专题命中 视频理解 ：video understanding(abstract)；video reasoning(abstract)；分类 cs.CV

AI总结 该研究提出一种无需训练的事后维纳表示滤波技术，通过离线校准校正视觉语言模型深层前馈输出投影，可在保持运行速度的同时降低其对象幻觉，在多类模型及基准上均验证了有效性与通用性。

GSTEP: Global Spatio-Temporal Density-Driven Visual Token Pruning for Efficient Video Large Language Models

GSTEP：面向高效视频大语言模型的全局时空密度驱动视觉令牌剪枝

Mengjie Zhang, Qihui Zhu, Tao Zhang, Shuangwu Chen, Huihuang Qin, Yu Guo, Shenghao Ye, Zijian Wen, Yunpeng Hou, Dong Jin, Xiaobin Tan, Huasen He, Jian Yang

专题命中 视频理解 ：video understanding(abstract)；long video(abstract)；分类 cs.CV

AI总结 本文提出即插即用的GSTEP剪枝框架，解决现有视频令牌剪枝方法的局部剪枝缺陷，在多个VideoLLMs上实现良好的准确率-效率权衡，在LLaVA-OneVision-7B上剪去75%视觉令牌，保留100.2%原始性能并获1.17倍端到端加速。

Comments 4 figures, accepted to ACM MM 26'

WaveZip: Wavelet-Driven Space-Time Decoupling for Video Token Condensation

WaveZip：用于视频令牌压缩的小波驱动时空解耦

Yuhui Zeng, Wang Chen, Jinfa Huang, Tianyu Xie, Yongdong Luo, Jiayi Ji, Xiawu Zheng, jiebo Luo

机构 * Media Analytics and Computing Lab, Xiamen University（厦门大学媒体分析与计算实验室） ； Department of Computer Science, University of Rochester（罗切斯特大学计算机科学系） 纠错

专题命中 视频理解 ：video understanding(abstract)；long video(abstract)；分类 cs.CV

AI总结 针对长视频理解中视觉令牌计算成本高的问题，提出WaveZip框架，利用离散小波变换分离时空信号，无需特定任务训练，能集成到LVLMs提升推理效率，在长视频理解基准测试中表现优异。

Comments 13 pages, 10 figures

HAS: Highlight-guided Attention Steering for Multimodal LLM Video Summarization

HAS：用于多模态大语言模型视频摘要的高亮引导注意力转向

Rui Chu, Yingjie Lao

机构 * Tufts University（塔夫茨大学） 纠错

专题命中 视频理解 ：video generation(abstract)；video understanding(abstract)；分类 cs.CV

AI总结 针对视频摘要，提出HAS方法，通过全局找连续帧级高亮分布并作为注意力转向向量，让多模态大语言模型在推理时更关注高亮帧，避免丢失信息，在多种基准测试中性能出色。

Track and Caption Any Motion: Open-Vocabulary Spatiotemporal Captioning via Trajectory-Conditioned Generation

跟踪并描述任何运动：通过轨迹条件生成实现开放词汇时空字幕

Bishoy Galoaa, Sarah Ostadabbas

机构 * Northeastern University（东北大学） 纠错

专题命中 视频理解 ：video understanding(abstract)；video-language(abstract)；分类 cs.CV

AI总结 研究提出TCAM框架，无需文本查询和区域提示，通过字幕感知重采样器在点粒度结合跟踪与语言，用现有分割注释训练，能描述视频中运动、定位时间及轨迹，优于密集视频字幕基线，匹配相关方法，为运动驱动视频理解提供新途径。