Papers with video understanding
O2NA: An Object-Oriented Non-Autoregressive Approach for Controllable Video Captioning (2021.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for video captioning consider a sequence of frames and biases towards focused objects. |
| Approach: | They propose an Object-Oriented Non-Autoregressive approach to video captioning . it performs three steps: 1) identify the focused objects and predict their locations . 2) generate related attribute words and relation words of these focused objects to form a draft caption . |
| Outcome: | The proposed method achieves competitive results with the state-of-the-art methods but with higher diversity and faster inference speed. |
Can CLIP Count Stars? An Empirical Study on Quantity Bias in CLIP (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Despite its versatility, CLIP-based applications often suffer from misunderstandings regarding user intent, leading to discrepancies between the required number of objects and the actual outputs. |
| Approach: | They empirically evaluate CLIP’s understanding of quantity from text, image, and cross-modal perspectives by carefully designing different experimental settings and datasets. |
| Outcome: | The proposed model has shown significant success in various downstream tasks, including editing, generation, and quality evaluation. |
Video2Commonsense: Generating Commonsense Descriptions to Enrich Video Captioning (2020.emnlp-main)
Copied to clipboard
| Challenge: | Observable changes in the scene are reflected in captions, but actions are also linked to social aspects such as intentions, effects, and attributes that describe the agent. |
| Approach: | They propose to generate captions from videos that describe latent aspects of the human agent's actions. |
| Outcome: | The proposed model can be used to describe latent aspects of human actions in video clips and answer questions about videos. |
See More, Store Less: Memory-Efficient Resolution for Video Moment Retrieval (2026.findings-eacl)
Copied to clipboard
| Challenge: | Existing video moment retrieval methods rely on sparse frame sampling, risking information loss. |
| Approach: | a new video-based framework enhances memory efficiency while maintaining high information resolution . SMORE uses query-guided captions to encode semantics aligned with user intent . |
| Outcome: | a new framework improves memory efficiency while maintaining high information resolution . it achieves state-of-the-art performance on QVHighlights, Charades-STA, and ActivityNet-Captions benchmarks . |
TUNA: Comprehensive Fine-grained Temporal Understanding Evaluation on Dense Dynamic Videos (2025.acl-long)
Copied to clipboard
Fanheng Kong, Jingyuan Zhang, Hongzhi Zhang, Shi Feng, Daling Wang, Linhao Yu, Xingguang Ji, Yu Tian, V. W., Fuzheng Zhang
| Challenge: | Existing benchmarks for video understanding often focus on specific aspects, overlooking the holistic nature of video content. |
| Approach: | They propose a temporal-oriented benchmark for fine-grained understanding on dense dynamic videos with two complementary tasks: captioning and QA. |
| Outcome: | The proposed model performs well on diverse video scenarios and dynamic videos, with interpretable and robust evaluation criteria. |
Video Compression Commander: Plug-and-Play Inference Acceleration for Video Large Language Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Recent studies have shown that Video Large Language Models (Vide-oLLMs) are efficient at video understanding but lack the quadratic complexity of visual tokens. |
| Approach: | They propose a plug-and-play inference acceleration framework for VideoLLM token compression that quantifies each frame’s uniqueness and adaptively adjusts compression intensity across frames. |
| Outcome: | Extensive experiments on video large language models and benchmarks show that the proposed framework can preserve essential information while reducing redundancy in video sequences. |
CAMEL-Bench: A Comprehensive Arabic LMM Benchmark (2025.findings-naacl)
Copied to clipboard
Sara Ghaboura, Ahmed Heakl, Omkar Thawakar, Ali Husain Salem Abdulla Alharthi, Ines Riahi, Abduljalil Radman, Jorma Laaksonen, Fahad Shahbaz Khan, Salman Khan, Rao Muhammad Anwer
| Challenge: | Recent years have witnessed a significant interest in developing large multimodal models capable of performing various visual reasoning and understanding tasks. |
| Approach: | They propose to use Arabic as a language to evaluate large multi-modal models capable of performing visual reasoning and understanding tasks. |
| Outcome: | The proposed benchmark comprises eight diverse domains and 38 sub-domains to represent a large population of over 400 million speakers. |
Watch, Listen, and Describe: Globally and Locally Aligned Cross-Modal Attentions for Video Captioning (N18-2)
Copied to clipboard
| Challenge: | Existing multi-modal fusion methods have shown encouraging results in video understanding, but how to selectively fuse the multi-dimensional representations at different levels of details remains unexplored. |
| Approach: | They propose a hierarchically aligned cross-modal attention framework to fuse audio and visual cues at different levels of detail. |
| Outcome: | The proposed framework outperforms the previous best systems on the video captioning task. |
Reasoning is All You Need for Video Generalization: A Counterfactual Benchmark with Sub-question Evaluation (2025.findings-acl)
Copied to clipboard
Qiji Zhou, YiFan Gong, Guangsheng Bao, Hongjie Qiu, Jinqiang Li, Xiangrong Zhu, Huajian Zhang, Yue Zhang
| Challenge: | Existing multimodal benchmarks often overlook counterfactual reasoning, which is crucial for robust video understanding. |
| Approach: | They propose a multidimensional multimodal benchmark that systematically evaluates MLLMs across the abstract-concrete and perception-cognition dimensions. |
| Outcome: | The proposed model decomposes complex queries into structured sub-questions, enabling fine-grained reasoning analysis. |
GRPO-CARE: Consistency-Aware Reinforcement Learning for Multimodal Reasoning (2026.findings-acl)
Copied to clipboard
| Challenge: | Recent reinforcement learning approaches have advanced reasoning in Large Language Models (LLMs), yet their adaptation to multimodal LLMs remains underexplored. |
| Approach: | They propose a reinforcement learning framework that eliminates KL penalties and rewards consistency . they propose GRPO-CARE, which outperforms standard GR PO, with a base reward for accuracy and an adaptive bonus for consistency. |
| Outcome: | The proposed framework outperforms standard GRPO on the most difficult evaluation level and reasoning consistency test benchmarks. |
Interesting Culture: Social Relation Recognition from Videos via Culture De-confounding (2025.findings-emnlp)
Copied to clipboard
| Challenge: | a culturally-specific cultural context can be used to train relationship recognition models . cultural confounding factors can be learned, limiting ability to recognize social relationships in different cultures. |
| Approach: | They propose a culturally-based model that mitigates the influence of culture . they also construct a video social relation recognition dataset to facilitate discussion . |
| Outcome: | The proposed model surpasses state-of-the-art methods on several datasets. |
InternLM-XComposer2.5-Reward: A Simple Yet Effective Multi-Modal Reward Model (2025.findings-acl)
Copied to clipboard
Yuhang Zang, Xiaoyi Dong, Pan Zhang, Yuhang Cao, Ziyu Liu, Shengyuan Ding, Shenxi Wu, Yubo Ma, Haodong Duan, Wenwei Zhang, Kai Chen, Dahua Lin, Jiaqi Wang
| Challenge: | Despite the promising performance of Large Vision Language Models, they sometimes generate incorrect outputs. |
| Approach: | They propose a multi-modal reward model that aligns LVLMs with human preferences. |
| Outcome: | The proposed model achieves excellent results on the latest multi-modal reward model benchmark and shows competitive performance on text-only reward model. |
VideoINSTA: Zero-shot Long Video Understanding via Informative Spatial-Temporal Reasoning with LLMs (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Long video understanding presents unique challenges due to the complexity of reasoning over extended timespans. |
| Approach: | They propose a framework VideoINSTA to leverage large language models for video understanding . they propose 'event-based temporalreasoning' and 'content-based spatial reasoning' |
| Outcome: | The proposed model significantly improves state-of-the-art on three long video question-answering benchmarks. |
VQAGuider: Guiding Multimodal Large Language Models to Answer Complex Video Questions (2025.acl-long)
Copied to clipboard
| Challenge: | Multimodal large language models (MLLMs) can grasp the intention of a question and decomposing it to a series of visual recognition sub-tasks to find out the answer with the help of an agent. |
| Approach: | They propose a framework for multimodal large language models to grasp the intention of a question and decompose it into a series of visual recognition sub-tasks to find out the answer. |
| Outcome: | The proposed framework improves the accuracy of complex video-related questions by 29.6% and 17.2% on CVQA and the existing VQA datasets. |
Learning to Focus on the Foreground for Temporal Sentence Grounding (2022.coling-1)
Copied to clipboard
| Challenge: | Existing methods for temporal sentence grounding do not capture subtle details of small objects. |
| Approach: | They propose a detection-free framework for temporal sentence grounding that learns to locate foreground regions related to the query in consecutive frames. |
| Outcome: | The proposed framework outperforms state-of-the-art methods on three challenging datasets. |
HAIC: Improving Human Action Understanding and Generation with Better Captions for Multi-modal Large Language Models (2025.acl-long)
Copied to clipboard
Xiao Wang, Jingyun Hua, Weihong Lin, Yuanxing Zhang, Fuzheng Zhang, Jianlong Wu, Di Zhang, Liqiang Nie
| Challenge: | Existing studies have shown that high-quality video captions can improve MLLMs' performance on videos involving human actions. |
| Approach: | They propose a data annotation pipeline to collect videos featuring clear human actions from the Internet and annotate them in a standardized caption format that uses human attributes to distinguish individuals. |
| Outcome: | The proposed pipeline combines two datasets to evaluate human action understanding. |
OmAgent: A Multi-modal Agent Framework for Complex Video Understanding with Task Divide-and-Conquer (2024.emnlp-main)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have expanded their capabilities to multimodal contexts, including comprehensive video understanding. |
| Approach: | They propose to store and retrieve relevant video frames for specific queries and a Divide-and-Conquer loop capable of autonomous reasoning. |
| Outcome: | The proposed model efficiently stores and retrieves relevant video frames for specific queries, preserving the detailed content of videos. |
D-CoDe: Scaling Image-Pretrained VLMs to Video via Dynamic Compression and Question Decomposition (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing training-free adaptation frameworks for video large language models are limited by the scarcity of high-quality video-text data. |
| Approach: | They propose a training-free adaptation framework that incorporates dynamic compression and question decomposition to address perception bottleneck and token overload. |
| Outcome: | The proposed framework improves video understanding across various benchmarks and on long-video benchmarks. |
A Video Is Worth 4096 Tokens: Verbalize Videos To Understand Them In Zero Shot (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing annotated training datasets hinder development of supervised learning models for multimedia content . lack of annotating benchmarks hinders development of models with satisfactory performance . a recent study shows that large language models have zero-shot performance in multimedia understanding . |
| Approach: | They propose to verbalize long videos to generate their descriptions in natural language . they then perform video-understanding tasks on the generated story as opposed to the original video . |
| Outcome: | The proposed method achieves better results than baselines for video understanding. |
AlanaVLM: A Multimodal Embodied AI Foundation Model for Egocentric Video Understanding (2024.findings-emnlp)
Copied to clipboard
Alessandro Suglia, Claudio Greco, Katie Baker, Jose Part, Ioannis Papaioannou, Arash Eshghi, Ioannis Konstas, Oliver Lemon
| Challenge: | Current Vision-Language Models (VLMs) focus on third-person view videos, neglecting the richness of egocentric perceptual experience. |
| Approach: | They propose to use the Egocentric Video Understanding Dataset (EVUD) to train VLMs on video captioning and question answering tasks specific to egocentric videos. |
| Outcome: | The proposed model outperforms open-source models including strong Socratic models using GPT-4 as a planner by 3.6% and outperformed Claude 3 and Gemini Pro Vision 1.0. |
VRoPE: Rotary Position Embedding for Video Large Language Models (2025.emnlp-main)
Copied to clipboard
Zikang Liu, Longteng Guo, Yepeng Tang, Tongtian Yue, Junxian Cai, Kai Ma, Qingbin Liu, Xi Chen, Jing Liu
| Challenge: | Existing versions of Large Language Models (LLMs) lack a positional encoding strategy for video. |
| Approach: | They propose a new positional encoding method tailored for Video-LLMs that mitigates positional biases and ensures a more uniform distribution of spatial focus. |
| Outcome: | The proposed method outperforms existing versions of RoPE in video understanding and reasoning tasks. |
Task-Specific Information Decomposition for End-to-End Dense Video Captioning (2025.acl-long)
Copied to clipboard
| Challenge: | Existing methods for label assignment fail to ensure relevance of localization information to descriptions. |
| Approach: | They propose a decomposed dense video captioning framework that derives localization and captioning queries from event queries, enabling task-specific representations while maintaining inter-task collaboration. |
| Outcome: | Experiments on YouCook2 and ActivityNet Captions show that the proposed framework achieves state-of-the-art performance. |
Captioning for Text-Video Retrieval via Dual-Group Direct Preference Optimization (2025.findings-emnlp)
Copied to clipboard
| Challenge: | auxiliary captions are generic and indistinguishable across visually similar videos . conventional captioning approaches are evaluated using language relevance scores . |
| Approach: | They propose a retrieval framework that directly optimizes caption generation using retrieval relevance scores. |
| Outcome: | The proposed retrieval framework optimizes caption generation using retrieval relevance scores . dual-group direct preference optimization is a learning strategy that supervises captioning . |
ViPE: Visual Perception in Parameter Space for Efficient Video-Language Understanding (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing video-language models rely on concatenating visual tokens with textual inputs for joint modeling, but this method suffers from significant inefficiency when scaling to long videos with dense visual inputs. |
| Approach: | They propose a video-to-parameter efficiency paradigm called ViPE that transforms video content into visual perceptual weights, which are directly injected into the LLM’s parameters. |
| Outcome: | The proposed model reduces FLOPs by 85% and inference time by up to 65% while reducing FLOP and FLOP inference times by up-to-65%. |
Exploring Response Uncertainty in MLLMs: An Empirical Evaluation under Misleading Scenarios (2025.emnlp-main)
Copied to clipboard
Yunkai Dang, Mengxi Gao, Yibo Yan, Xin Zou, Yanggan Gu, Jungang Li, Jingyu Wang, Peijie Jiang, Aiwei Liu, Jia Liu, Xuming Hu
| Challenge: | Existing studies have focused mainly on visual–textual misalignment, leaving largely unexplored the MLLMs’ ability to preserve an original correct answer when confronted with misleading information. |
| Approach: | They propose a two-stage evaluation pipeline to quantify the response uncertainty phenomenon by eliciting each model’s original response on unperturbed inputs and injecting explicit (false-answer hints) and implicit (contextual contradictions) misleading instructions. |
| Outcome: | The proposed model overturns a correct answer in 65% of cases after receiving a single deceptive cue. |
VideoCuRL: Video Curriculum Reinforcement Learning with Orthogonal Difficulty Decomposition (2026.acl-long)
Copied to clipboard
| Challenge: | Reinforcement Learning (RL) is crucial for Video-LLMs with complex spatiotemporal reasoning. |
| Approach: | They propose a framework that decomposes difficulty into two axes in video understanding . they employ efficient, training-free proxies to map data onto a 2D curriculum grid . |
| Outcome: | The proposed framework surpasses strong RL baselines on reasoning and perception tasks. |
MVTamperBench: Evaluating Robustness of Vision-Language Models (2025.findings-acl)
Copied to clipboard
Amit Agarwal, Srikant Panda, Angeline Charles, Hitesh Laxmichand Patel, Bhargava Kumar, Priyaranjan Pattnayak, Taki Hasan Rafi, Tejaswini Kumar, Hansa Meghwani, Karan Gupta, Dong-Kyu Chae
| Challenge: | Multimodal Large Language Models (MLLMs) have been a key advance in video understanding but their vulnerability to adversarial tampering remains underexplored. |
| Approach: | They evaluate MLLMs against five prevalent tampering techniques to assess their robustness . they use a tampered video format to examine the vulnerability of ML models . |
| Outcome: | The benchmark evaluates MLLMs against five prevalent tampering techniques based on 19 video manipulation tasks. |
InfiniBench: A Benchmark for Large Multi-Modal Models in Long-Form Movies and TV Shows (2025.emnlp-main)
Copied to clipboard
Kirolos Ataallah, Eslam Mohamed Bakr, Mahmoud Ahmed, Chenhui Gou, Khushbu Pahwa, Jian Ding, Mohamed Elhoseiny
| Challenge: | Existing benchmarks fail to test the full range of cognitive skills needed to process long-form videos . |
| Approach: | They propose a benchmark to evaluate models' ability to process long-form videos rigorously. |
| Outcome: | The benchmark measures the cognitive skills of models in understanding long-form videos . it offers the largest set of question-answer pairs for long video comprehension . |
Video-Skill-CoT: Skill-based Chain-of-Thoughts for Domain-Adaptive Video Reasoning (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for chain-of-thought reasoning fail to adapt to domain-specific skills over video content. |
| Approach: | They propose a framework that automatically constructs and leverages skill-aware CoT supervisions for domain-adaptive video reasoning. |
| Outcome: | The proposed framework outperforms strong baselines on three video understanding benchmarks. |
MaCP: Minimal yet Mighty Adaptation via Hierarchical Cosine Projection (2025.acl-long)
Copied to clipboard
| Challenge: | MaCP is a new adaptation method for large foundation models that requires minimal parameters and memory for fine-tuning. |
| Approach: | They propose a method that exploits the superior energy compaction and decorrelation properties of cosine projection to improve model efficiency and accuracy. |
| Outcome: | The proposed method improves model efficiency and accuracy across a wide range of single-modality tasks including natural language understanding, natural language generation, text summarization, and multi-modalities such as image classification and video understanding. |
A Culturally-diverse Multilingual Multimodal Video Benchmark & Model (2025.emnlp-main)
Copied to clipboard
Bhuiyan Sanjid Shafique, Ashmal Vayani, Muhammad Maaz, Hanoona Abdul Rasheed, Dinura Dissanayake, Mohammed Irfan Kurpath, Yahya Hmaiti, Go Inoue, Jean Lahoud, Md. Safirur Rashid, Shadid Intisar Quasem, Maheen Fatima, Franco Vidal, Mykola Maslych, Ketan Pravin More, Sanoojan Baliah, Hasindri Watawana, Yuhao Li, Fabian Farestam, Leon Schaller, Roman Tymtsiv, Simon Weber, Hisham Cholakkal, Ivan Laptev, Shin’ichi Satoh, Michael Felsberg, Mubarak Shah, Salman Khan, Fahad Shahbaz Khan
| Challenge: | Large multimodal models have gained attention for their effectiveness to understand and generate descriptions of visual content. |
| Approach: | They propose a multilingual Video LMM benchmark to evaluate video LMMs across 14 languages . they also introduce a machine translated multilingual video training set . |
| Outcome: | The proposed video LMM benchmark is designed to evaluate video Lmms across 14 languages including Arabic, Bengali, Chinese, English, French, German, Hindi, Japanese, Russian, Sinhala, Spanish, Swedish, Tamil, and Urdu. |
TV-TREES: Multimodal Entailment Trees for Neuro-Symbolic Video Reasoning (2024.emnlp-main)
Copied to clipboard
| Challenge: | TV-TREES is the first multimodal entailment tree generator for video understanding . it searches for trees of enanglement relationships between text-video evidence and higher-level conclusions that prove question-answer pairs. |
| Approach: | They propose a multimodal entailment tree generator that promotes interpretable joint-modality reasoning by searching for trees of enanglement relationships between simple text-video evidence and higher-level conclusions that prove question-answer pairs. |
| Outcome: | The proposed approach performs on the TVQA benchmark and shows that it is state-of-the-art on full clips. |
See the Forest for the Trees: Loosely Speculative Decoding via Visual-Semantic Guidance for Efficient Inference of Video LLMs (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods for video understanding suffer from autoregressive generation of tokens. |
| Approach: | They propose a training-free loosely SD framework for Video-LLMs that uses visual-relevant tokens to accurately pinpoint the latter. |
| Outcome: | The proposed framework boosts the accepted length and speedup ratio by 136% and 35% compared to SOTA training-free SD methods for Video-LLMs. |
LongLLaVA: Scaling Multi-modal LLMs to 1000 Images Efficiently via a Hybrid Architecture (2025.findings-emnlp)
Copied to clipboard
Xidong Wang, Dingjie Song, Shunian Chen, Junying Chen, Zhenyang Cai, Chen Zhang, Lichao Sun, Benyou Wang
| Challenge: | Long-context Large Language Models (MLLMs) are critical for video understanding and image analysis. |
| Approach: | They propose a hybrid architecture that integrates Mamba and Transformer blocks . they introduce data construction methods that capture both temporal and spatial dependencies . |
| Outcome: | The proposed model achieves competitive results across various benchmarks while maintaining high throughput and low memory consumption. |
GLIMPSE: Do Large Vision-Language Models Truly Think With Videos or Just Glimpse at Them? (2025.emnlp-main)
Copied to clipboard
Yiyang Zhou, Linjie Li, Shi Qiu, Zhengyuan Yang, Yuyang Zhao, Siwei Han, Yangfan He, Kangqi Li, Haonian Ji, Zihao Zhao, Haibo Tong, Lijuan Wang, Huaxiu Yao
| Challenge: | Existing video benchmarks often resemble image-based questions with scans of only a few key frames, without deep temporal reasoning. |
| Approach: | They propose a video benchmark to assess whether large vision-language models can genuinely think with videos rather than perform superficial frame-level analysis. |
| Outcome: | The proposed benchmark consists of 3,269 videos and over 4,342 highly visual-centric questions across 11 categories, including Trajectory Analysis, Temporal Reasoning, and Forensics Detection. |
GameplayQA: A Benchmarking Framework for Decision-Dense POV-Synced Multi-Video Understanding of 3D Virtual Agents (2026.acl-long)
Copied to clipboard
Yunzhe Wang, Runhui Xu, Kexin Zheng, Tianyi Zhang, Jayavibhav Niranjan Kogundi, Soham Hans, Volkan Ustun
| Challenge: | Multimodal Large Language Models are emerging as a backbone for autonomous agents in 3D environments. |
| Approach: | They propose a framework for evaluating agentic-centric perception and reasoning through video understanding. |
| Outcome: | The proposed framework evaluates agentic-centric perception and reasoning through video understanding. |