Video_Caption
-
Chain-of-Glimpse: Search-Guided Progressive Object-Grounded Reasoning for Video Understanding
2026-04-16 06:50:20Zhixuan Wu, Quanxing Zha, Teng Wang, Genbao Xu, Wenyuan Gu, Wei Rao, Nan Ma, Bo Cheng, Soujanya PoriaAbstract
Video understanding requires identifying and reasoning over semantically discriminative visual objects across frames, yet existing object-agnostic solutions struggle to effectively handle substantial object variations over time. To address this, we introduce Chain-of-Glimpse, a search-guided progressive object-grounded reasoning framework that explicitly anchors each reasoning step to specific visual evidence regions, enabling compositional and multi-step decision-making. Formally, Chain-of-Glimpse formulates video reasoning as a step-by-step process that incrementally builds spatially grounded traces around task-relevant visual objects, thereby mitigating over-reliance on saliency-driven cues. Specifically, Chain-of-Glimpse features a search-guided controller, optimized via reinforcement learning with a format reward that significantly incentivizes grounding capability, to iteratively ground visual evidence regions and form reliable reasoning trajectories, yielding accurate and interpretable multi-step decisions. Extensive evaluations on both in domain NExTQA and out-of-domain Video-Holmes, CG-Bench Reasoning, and VRBench benchmarks demonstrate consistent performance gains, robustness and generalization of Chain-of-Glimpse across diverse video reasoning tasks.
Abstract (translated)
URL
https://arxiv.org/abs/2604.14692
PDF
-
One Token per Highly Selective Frame: Towards Extreme Compression for Long Video Understanding
2026-04-15 17:59:52Zheyu Zhang, Ziqi Pang, Shixing Chen, Xiang Hao, Vimal Bhat, Yu-Xiong WangAbstract
Long video understanding is inherently challenging for vision-language models (VLMs) because of the extensive number of frames. With each video frame typically expanding into tens or hundreds of tokens, the limited context length of large language models (LLMs) forces the VLMs to perceive the frames sparsely and lose temporal information. To address this, we explore extreme video token compression towards \emph{one token per frame} at the final LLM layer. Our key insight is that heuristic-based compression, widely adopted by previous methods, is prone to information loss, and this necessitates supervising LLM layers into \emph{learnable} and \emph{progressive} modules for \emph{token-level compression} (LP-Comp). Such compression enables our VLM to digest 2x-4x more frames with improved performance. To further increase the token efficiency, we investigate \emph{frame-level compression}, which selects the frames most relevant to the queries via the internal attention scores of the LLM layers, named \emph{question-conditioned compression} (QC-Comp). As a notable distinction from previous studies, we mitigate the position bias of LLM attention in long contexts, \emph{i.e.}, the over-concentration on the beginning and end of a sequence, by splitting long videos into short segments and employing local attention. Collectively, our combined \emph{token-level} and \emph{frame-level} leads to an e\textbf{x}treme compression model for long video understanding, named \textbf{\name}, achieving a significantly larger compression ratio and enabling denser frame sampling. Our \name is finetuned from VideoChat-Flash with a data-efficient \emph{supervised compression tuning} stage that only requires 2.5\% of the supervised fine-tuning data, yet boosts the accuracy from 42.9\% to 46.2\% on LVBench and enhances multiple other long video benchmarks.
Abstract (translated)
URL
https://arxiv.org/abs/2604.14149
PDF
-
Relaxing Anchor-Frame Dominance for Mitigating Hallucinations in Video Large Language Models
2026-04-14 11:05:42Zijian Liu, Sihan Cao, Pengcheng Zheng, Kuien Liu, Caiyan Qin, Xiaolin Qin, Jiwei Wei, Chaoning ZhangAbstract
Recent Video Large Language Models (Video-LLMs) have demonstrated strong capability in video understanding, yet they still suffer from hallucinations. Existing mitigation methods typically rely on training, input modification, auxiliary guidance, or additional decoding procedures, while largely overlooking a more fundamental challenge. During generation, Video-LLMs tend to over-rely on a limited portion of temporal evidence, leading to temporally imbalanced evidence aggregation across the video. To address this issue, we investigate a decoder-side phenomenon in which the model exhibits a temporally imbalanced concentration pattern. We term the frame with the highest aggregated frame-level attention mass the anchor frame. We find that this bias is largely independent of the input video and instead appears to reflect a persistent, model-specific structural or positional bias, whose over-dominance is closely associated with hallucination-prone generation. Motivated by this insight, we propose Decoder-side Temporal Rebalancing (DTR), a training-free, layer-selective inference method that rebalances temporal evidence allocation in middle-to-late decoder layers without altering visual encoding or requiring auxiliary models. DTR adaptively calibrates decoder-side visual attention to alleviate temporally imbalanced concentration and encourage under-attended frames to contribute more effectively to response generation. In this way, DTR guides the decoder to ground its outputs in temporally broader and more balanced video evidence. Extensive experiments on hallucination and video understanding benchmarks show that DTR consistently improves hallucination robustness across diverse Video-LLM families, while preserving competitive video understanding performance and high inference efficiency.
Abstract (translated)
URL
https://arxiv.org/abs/2604.12582
PDF
-
Unlocking the Potential of Grounding DINO in Videos: Parameter-Efficient Adaptation for Limited-Data Spatial-Temporal Localization
2026-04-14 06:32:24Zanyi Wang, Fan Li, Dengyang Jiang, Liuzhuozheng Li, Yunhua Zhong, Guang Dai, Mengmeng WangAbstract
Spatio-temporal video grounding (STVG) aims to localize queried objects within dynamic video segments. Prevailing fully-trained approaches are notoriously data-hungry. However, gathering large-scale STVG data is exceptionally challenging: dense frame-level bounding boxes and complex temporal language alignments are prohibitively expensive to annotate, especially for specialized video domains. Consequently, conventional models suffer from severe overfitting on these inherently limited datasets, while zero-shot foundational models lack the task-specific temporal awareness needed for precise localization. To resolve this small-data challenge, we introduce ST-GD, a data-efficient framework that adapts pre-trained 2D visual-language models (e.g., Grounding DINO) to video tasks. To avoid destroying pre-trained priors on small datasets, ST-GD keeps the base model frozen and strategically injects lightweight adapters (~10M trainable parameters) to instill spatio-temporal awareness, alongside a novel temporal decoder for boundary prediction. This design naturally counters data scarcity. Consequently, ST-GD excels in data-scarce scenarios, achieving highly competitive performance on the limited-scale HC-STVG v1/v2 benchmarks, while maintaining robust generalization on the VidSTG dataset. This validates ST-GD as a powerful paradigm for complex video understanding under strict small-data constraints.
Abstract (translated)
URL
https://arxiv.org/abs/2604.12346
PDF
-
All in One: A Unified Synthetic Data Pipeline for Multimodal Video Understanding
2026-04-14 06:17:35Tanzila Rahman, Renjie Liao, Leonid SigalAbstract
Training multimodal large language models (MLLMs) for video understanding requires large-scale annotated data spanning diverse tasks such as object counting, question answering, and segmentation. However, collecting and annotating multimodal video data in real-world is costly, slow, and inherently limited in diversity and coverage. To address this challenge, we propose a unified synthetic data generation pipeline capable of automatically producing unlimited multimodal video data with rich and diverse supervision. Our framework supports multiple task formats within a single pipeline, enabling scalable and consistent data creation across tasks. To further enhance reasoning ability, we introduce a VQA-based fine-tuning strategy that trains models to answer structured questions about visual content rather than relying solely on captions or simple instructions. This formulation encourages deeper visual grounding and reasoning. We evaluate our approach in three challenging tasks: video object counting, video-based visual question answering, and video object segmentation. Experimental results demonstrate that models trained predominantly on synthetic data generalize effectively to real-world datasets, often outperforming traditionally trained counterparts. Our findings highlight the potential of unified synthetic data pipelines as a scalable alternative to expensive real-world annotation for multimodal video understanding.
Abstract (translated)
URL
https://arxiv.org/abs/2604.12335
PDF
-
ViLL-E: Video LLM Embeddings for Retrieval
2026-04-13 23:54:58Rohit Gupta, Jayakrishnan Unnikrishnan, Fan Fei, Sheng Liu, Son Tran, Mubarak ShahAbstract
Video Large Language Models (VideoLLMs) excel at video understanding tasks where outputs are textual, such as Video Question Answering and Video Captioning. However, they underperform specialized embedding-based models in Retrieval tasks, such as Text-toVideo Retrieval and Moment Retrieval. We introduce ViLL-E (Video-LLM-Embed), a unified VideoLLM architecture endowed with a novel embedding generation mechanism that allows the model to "think longer" for complex videos and stop early for easy ones. We train this model with a three-stage training methodology combining generative and contrastive learning: initial large-scale pre-training with video-caption pairs; followed by continual training on a smaller, detailed-caption dataset; and concluding with task-specific fine-tuning on a novel multi-task dataset covering Video QA, Temporal Localization, Video Retrieval, and Video-Text Matching. Our model significantly improves temporal localization (on avg. 7% over other VideoLLMs) and video retrieval (up to 4% over dual encoder models), achieving performance comparable to state-of-the-art specialized embedding models while remaining competitive on VideoQA tasks. Furthermore, our joint contrastive-generative training unlocks new zero-shot capabilities, significantly outperforming state-of-the-art methods in composed video retrieval (+5% over SotA) and retrieval from long text (+2% over SotA).
Abstract (translated)
URL
https://arxiv.org/abs/2604.12148
PDF
-
STS-Mixer: Spatio-Temporal-Spectral Mixer for 4D Point Cloud Video Understanding
2026-04-13 15:47:23Wenhao Li, Xueying Jiang, Gongjie Zhang, Xiaoqin Zhang, Ling Shao, Shijian LuAbstract
4D point cloud videos capture rich spatial and temporal dynamics of scenes which possess unique values in various 4D understanding tasks. However, most existing methods work in the spatiotemporal domain where the underlying geometric characteristics of 4D point cloud videos are hard to capture, leading to degraded representation learning and understanding of 4D point cloud videos. We address the above challenge from a complementary spectral perspective. By transforming 4D point cloud videos into graph spectral signals, we can decompose them into multiple frequency bands each of which captures distinct geometric structures of point cloud videos. Our spectral analysis reveals that the decomposed low-frequency signals capture more coarse shapes while high-frequency signals encode more fine-grained geometry details. Building on these observations, we design Spatio-Temporal-Spectral Mixer (STS-Mixer), a unified framework that mixes spatial, temporal, and spectral representations of point cloud videos. STS-Mixer integrates multi-band delineated spectral signals with spatiotemporal information to capture rich geometries and temporal dynamics, while enabling fine-grained and holistic understanding of 4D point cloud videos. Extensive experiments show that STS-Mixer achieves superior performance consistently across multiple widely adopted benchmarks on both 3D action recognition and 4D semantic segmentation tasks. Code and models are available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2604.11637
PDF
-
Empowering Video Translation using Multimodal Large Language Models
2026-04-13 10:42:31Bingzheng QU, Kehai Chen, Xuefeng Bai, Min ZhangAbstract
Recent developments in video translation have further enhanced cross-lingual access to video content, with multimodal large language models (MLLMs) playing an increasingly important supporting role. With strong multimodal understanding, reasoning, and generation capabilities, MLLMs-based video translation systems are overcoming the limitations of traditional cascaded pipelines that separately handle automatic speech recognition, machine translation, text-to-speech and lip synchronization. These MLLM-powered approaches not only achieve competitive or superior translation quality, but also demonstrate stronger robustness in zero-shot settings and multi-speaker scenarios, while jointly modeling semantic fidelity, timing, speaker identity, and emotional consistency. However, despite the rapid progress of MLLMs and extensive surveys on general video-language understanding, a focused and systematic review of how MLLMs empower video translation tasks is still lacking. To fill this gap, we provide the first comprehensive overview of MLLMs-based video translation, organized around a three-role taxonomy: 1) Semantic Reasoner, which characterizes how MLLMs perform video understanding, temporal reasoning, and multimodal fusion; 2) Expressive Performer, which analyzes LLM-driven and LLM-augmented techniques for expressive, controllable speech generation; and 3) Visual Synthesizer, which examines different types of video generators for high-fidelity lip-sync and visual alignment. Finally, we discuss open challenges in video understanding, temporal modeling, and multimodal alignment, and outline promising future research directions for MLLMs-powered video translation.
Abstract (translated)
URL
https://arxiv.org/abs/2604.11283
PDF
-
Script-a-Video: Deep Structured Audio-visual Captions via Factorized Streams and Relational Grounding
2026-04-13 09:50:36Tencent Hunyuan TeamAbstract
Advances in Multimodal Large Language Models (MLLMs) are transforming video captioning from a descriptive endpoint into a semantic interface for both video understanding and generation. However, the dominant paradigm still casts videos as monolithic narrative paragraphs that entangle visual, auditory, and identity information. This dense coupling not only compromises representational fidelity but also limits scalability, since even local edits can trigger global rewrites. To address this structural bottleneck, we propose Multi-Stream Scene Script (MTSS), a novel paradigm that replaces monolithic text with factorized and explicitly grounded scene descriptions. MTSS is built on two core principles: Stream Factorization, which decouples a video into complementary streams (Reference, Shot, Event, and Global), and Relational Grounding, which reconnects these isolated streams through explicit identity and temporal links to maintain holistic video consistency. Extensive experiments demonstrate that MTSS consistently enhances video understanding across various models, achieving an average reduction of 25% in the total error rate on Video-SALMONN-2 and an average performance gain of 67% on the Daily-Omni reasoning benchmark. It also narrows the performance gap between smaller and larger MLLMs, indicating a substantially more learnable caption interface. Finally, even without architectural adaptation, replacing monolithic prompts with MTSS in multi-shot video generation yields substantial human-rated improvements: a 45% boost in cross-shot identity consistency, a 56% boost in audio-visual alignment, and a 71% boost in temporal controllability.
Abstract (translated)
URL
https://arxiv.org/abs/2604.11244
PDF
-
OmniScript: Towards Audio-Visual Script Generation for Long-Form Cinematic Video
2026-04-13 07:19:27Junfu Pu, Yuxin Chen, Teng Wang, Ying ShanAbstract
Current multimodal large language models (MLLMs) have demonstrated remarkable capabilities in short-form video understanding, yet translating long-form cinematic videos into detailed, temporally grounded scripts remains a significant challenge. This paper introduces the novel video-to-script (V2S) task, aiming to generate hierarchical, scene-by-scene scripts encompassing character actions, dialogues, expressions, and audio cues. To facilitate this, we construct a first-of-its-kind human-annotated benchmark and propose a temporally-aware hierarchical evaluation framework. Furthermore, we present OmniScript, an 8B-parameter omni-modal (audio-visual) language model tailored for long-form narrative comprehension. OmniScript is trained via a progressive pipeline that leverages chain-of-thought supervised fine-tuning for plot and character reasoning, followed by reinforcement learning using temporally segmented rewards. Extensive experiments demonstrate that despite its parameter efficiency, OmniScript significantly outperforms larger open-source models and achieves performance comparable to state-of-the-art proprietary models, including Gemini 3-Pro, in both temporal localization and multi-field semantic accuracy.
Abstract (translated)
URL
https://arxiv.org/abs/2604.11102
PDF
-
GTASA: Ground Truth Annotations for Spatiotemporal Analysis, Evaluation and Training of Video Models
2026-04-12 00:01:51Nicolae Cudlenco, Mihai Masala, Marius LeordeanuAbstract
Generating complex multi-actor scenario videos remains difficult even for state-of-the-art neural generators, while evaluating them is hard due to the lack of ground truth for physical plausibility and semantic faithfulness. We introduce GTASA, a corpus of multi-actor videos with per-frame spatial relation graphs and event-level temporal mappings, and the system that produced it based on Graphs of Events in Space and Time (GEST): GEST-Engine. We compare our method with both open and closed source neural generators and prove both qualitatively (human evaluation of physical validity and semantic alignment) and quantitatively (via training video captioning models) the clear advantages of our method. Probing four frozen video encoders across 11 spatiotemporal reasoning tasks enabled by GTASA's exact 3D ground truth reveals that self-supervised encoders encode spatial structure significantly better than VLM visual encoders.
Abstract (translated)
URL
https://arxiv.org/abs/2604.10385
PDF
-
Demographic and Linguistic Bias Evaluation in Omnimodal Language Models
2026-04-11 03:58:02Alaa ElobaidAbstract
This paper provides a comprehensive evaluation of demographic and linguistic biases in omnimodal language models that process text, images, audio, and video within a single framework. Although these models are being widely deployed, their performance across different demographic groups and modalities is not well studied. Four omnimodal models are evaluated on tasks that include demographic attribute estimation, identity verification, activity recognition, multilingual speech transcription, and language identification. Accuracy differences are measured across age, gender, skin tone, language, and country of origin. The results show that image and video understanding tasks generally exhibit better performance with smaller demographic disparities. In contrast, audio understanding tasks exhibit significantly lower performance and substantial bias, including large accuracy differences across age groups, genders, and languages, and frequent prediction collapse toward narrow categories. These findings highlight the importance of evaluating fairness across all supported modalities as omnimodal language models are increasingly used in real-world applications.
Abstract (translated)
URL
https://arxiv.org/abs/2604.10014
PDF
-
Tango: Taming Visual Signals for Efficient Video Large Language Models
2026-04-10 17:59:56Shukang Yin, Sirui Zhao, Hanchao Wang, Baozhi Jia, Xianquan Wang, Chaoyou Fu, Enhong ChenAbstract
Token pruning has emerged as a mainstream approach for developing efficient Video Large Language Models (Video LLMs). This work revisits and advances the two predominant token-pruning paradigms: attention-based selection and similarity-based clustering. Our study reveals two critical limitations in existing methods: (1) conventional top-k selection strategies fail to fully account for the attention distribution, which is often spatially multi-modal and long-tailed in magnitude; and (2) direct similarity-based clustering frequently generates fragmented clusters, resulting in distorted representations after pooling. To address these bottlenecks, we propose Tango, a novel framework designed to optimize the utilization of visual signals. Tango integrates a diversity-driven strategy to enhance attention-based token selection, and introduces Spatio-temporal Rotary Position Embedding (ST-RoPE) to preserve geometric structure via locality priors. Comprehensive experiments across various Video LLMs and video understanding benchmarks demonstrate the effectiveness and generalizability of our approach. Notably, when retaining only 10% of the video tokens, Tango preserves 98.9% of the original performance on LLaVA-OV while delivering a 1.88x inference speedup.
Abstract (translated)
URL
https://arxiv.org/abs/2604.09547
PDF
-
Efficient Spatial-Temporal Focal Adapter with SSM for Temporal Action Detection
2026-04-10 09:51:15Yicheng Qiu, Keiji YanaiAbstract
Temporal human action detection aims to identify and localize action segments within untrimmed videos, serving as a pivotal task in video understanding. Despite the progress achieved by prior architectures like CNN and Transformer models, these continue to struggle with feature redundancy and degraded global dependency modeling capabilities when applied to long video sequences. These limitations severely constrain their scalability in real-world video analysis. State Space Models (SSMs) offer a promising alternative with linear long-term modeling and robust global temporal reasoning capabilities. Rethinking the application of SSMs in temporal modeling, this research constructs a novel framework for video human action detection. Specifically, we introduce the Efficient Spatial-Temporal Focal (ESTF) Adapter into the pre-trained layers. This module integrates the advantages of our proposed Temporal Boundary-aware SSM(TB-SSM) for temporal feature modeling with efficient processing of spatial features. We perform comprehensive and quantitative analyses across multiple benchmarks, comparing our proposed method against previous SSM-based and other structural methods. Extensive experiments demonstrate that our improved strategy significantly enhances both localization performance and robustness, validating the effectiveness of our proposed method.
Abstract (translated)
URL
https://arxiv.org/abs/2604.09164
PDF
-
StreamMeCo: Long-Term Agent Memory Compression for Efficient Streaming Video Understanding
2026-04-10 06:11:34Junxi Wang, Te Sun, Jiayi Zhu, Junxian Li, Haowen Xu, Zichen Wen, Xuming Hu, Zhiyu Li, Linfeng ZhangAbstract
Vision agent memory has shown remarkable effectiveness in streaming video understanding. However, storing such memory for videos incurs substantial memory overhead, leading to high costs in both storage and computation. To address this issue, we propose StreamMeCo, an efficient Stream Agent Memory Compression framework. Specifically, based on the connectivity of the memory graph, StreamMeCo introduces edge-free minmax sampling for the isolated nodes and an edge-aware weight pruning for connected nodes, evicting the redundant memory nodes while maintaining the accuracy. In addition, we introduce a time-decay memory retrieval mechanism to further eliminate the performance degradation caused by memory compression. Extensive experiments on three challenging benchmark datasets (M3-Bench-robot, M3-Bench-web and Video-MME-Long) demonstrate that under 70% memory graph compression, StreamMeCo achieves a 1.87* speedup in memory retrieval while delivering an average accuracy improvement of 1.0%. Our code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2604.09000
PDF
-
Small Vision-Language Models are Smart Compressors for Long Video Understanding
2026-04-09 11:40:25Junjie Fei, Jun Chen, Zechun Liu, Yunyang Xiong, Chong Zhou, Wei Wen, Junlin Han, Mingchen Zhuge, Saksham Suri, Qi Qian, Shuming Liu, Lemeng Wu, Raghuraman Krishnamoorthi, Vikas Chandra, Mohamed Elhoseiny, Chenchen ZhuAbstract
Adapting Multimodal Large Language Models (MLLMs) for hour-long videos is bottlenecked by context limits. Dense visual streams saturate token budgets and exacerbate the lost-in-the-middle phenomenon. Existing heuristics, like sparse sampling or uniform pooling, blindly sacrifice fidelity by discarding decisive moments and wasting bandwidth on irrelevant backgrounds. We propose Tempo, an efficient query-aware framework compressing long videos for downstream understanding. Tempo leverages a Small Vision-Language Model (SVLM) as a local temporal compressor, casting token reduction as an early cross-modal distillation process to generate compact, intent-aligned representations in a single forward pass. To enforce strict budgets without breaking causality, we introduce Adaptive Token Allocation (ATA). Exploiting the SVLM's zero-shot relevance prior and semantic front-loading, ATA acts as a training-free $O(1)$ dynamic router. It allocates dense bandwidth to query-critical segments while compressing redundancies into minimal temporal anchors to maintain the global storyline. Extensive experiments show our 6B architecture achieves state-of-the-art performance with aggressive dynamic compression (0.5-16 tokens/frame). On the extreme-long LVBench (4101s), Tempo scores 52.3 under a strict 8K visual budget, outperforming GPT-4o and Gemini 1.5 Pro. Scaling to 2048 frames reaches 53.7. Crucially, Tempo compresses hour-long videos substantially below theoretical limits, proving true long-form video understanding relies on intent-driven efficiency rather than greedily padded context windows.
Abstract (translated)
URL
https://arxiv.org/abs/2604.08120
PDF
-
DiffVC: A Non-autoregressive Framework Based on Diffusion Model for Video Captioning
2026-04-09 10:56:49Junbo Wang, Liangyu Fu, Yuke Li, Yining Zhu, Ya Jing, Xuecheng Wu, Jiangbin ZhengAbstract
Current video captioning methods usually use an encoder-decoder structure to generate text autoregressively. However, autoregressive methods have inherent limitations such as slow generation speed and large cumulative error. Furthermore, the few non-autoregressive counterparts suffer from deficiencies in generation quality due to the lack of sufficient multimodal interaction modeling. Therefore, we propose a non-autoregressive framework based on Diffusion model for Video Captioning (DiffVC) to address these issues. Its parallel decoding can effectively solve the problems of generation speed and cumulative error. At the same time, our proposed discriminative conditional Diffusion Model can generate higher-quality textual descriptions. Specifically, we first encode the video into a visual representation. During training, Gaussian noise is added to the textual representation of the ground-truth caption. Then, a new textual representation is generated via the discriminative denoiser with the visual representation as a conditional constraint. Finally, we input the new textual representation into a non-autoregressive language model to generate captions. During inference, we directly sample noise from the Gaussian distribution for generation. Experiments on MSVD, MSR-VTT, and VATEX show that our method can outperform previous non-autoregressive methods and achieve comparable performance to autoregressive methods, e.g., it achieved a maximum improvement of 9.9 on the CIDEr and improvement of 2.6 on the B@4, while having faster generation speed. The source code will be available soon.
Abstract (translated)
URL
https://arxiv.org/abs/2604.08084
PDF
-
AdaSpark: Adaptive Sparsity for Efficient Long-Video Understanding
2026-04-09 10:48:32Handong Li, Zikang Liu, Longteng Guo, Tongtian Yue, Yepeng Tang, Xinxin Zhu, Chuanyang Zheng, Ziming Wang, Zhibin Wang, Jun Song, Cheng Yu, Bo Zheng, Jing LiuAbstract
Processing long-form videos with Video Large Language Models (Video-LLMs) is computationally prohibitive. Current efficiency methods often compromise fine-grained perception through irreversible information disposal or inhibit long-range temporal modeling via rigid, predefined sparse patterns. This paper introduces AdaSpark, an adaptive sparsity framework designed to address these limitations. AdaSpark first partitions video inputs into 3D spatio-temporal cubes. It then employs two co-designed, context-aware components: (1) Adaptive Cube-Selective Attention (AdaS-Attn), which adaptively selects a subset of relevant video cubes to attend for each query token, and (2) Adaptive Token-Selective FFN (AdaS-FFN), which selectively processes only the most salient tokens within each cube. An entropy-based (Top-p) selection mechanism adaptively allocates computational resources based on input complexity. Experiments demonstrate that AdaSpark significantly reduces computational load by up to 57% FLOPs while maintaining comparable performance to dense models and preserving fine-grained, long-range dependencies, as validated on challenging hour-scale video benchmarks.
Abstract (translated)
URL
https://arxiv.org/abs/2604.08077
PDF
-
ABMAMBA: Multimodal Large Language Model with Aligned Hierarchical Bidirectional Scan for Efficient Video Captioning
2026-04-09 09:58:56Daichi Yashima, Shuhei Kurita, Yusuke Oda, Shuntaro Suzuki, Seitaro Otsuki, Komei SugiuraAbstract
In this study, we focus on video captioning by fully open multimodal large language models (MLLMs). The comprehension of visual sequences is challenging because of their intricate temporal dependencies and substantial sequence length. The core attention mechanisms of existing Transformer-based approaches scale quadratically with the sequence length, making them computationally prohibitive. To address these limitations, we propose Aligned Hierarchical Bidirectional Scan Mamba (ABMamba), a fully open MLLM with linear computational complexity that enables the scalable processing of video sequences. ABMamba extends Deep State Space Models as its language backbone, replacing the costly quadratic attention mechanisms, and employs a novel Aligned Hierarchical Bidirectional Scan module that processes videos across multiple temporal resolutions. On standard video captioning benchmarks such as VATEX and MSR-VTT, ABMamba demonstrates competitive performance compared to typical MLLMs while achieving approximately three times higher throughput.
Abstract (translated)
URL
https://arxiv.org/abs/2604.08050
PDF
-
Bridging Time and Space: Decoupled Spatio-Temporal Alignment for Video Grounding
2026-04-09 09:14:00Xuezhen Tu, Jingyu Wu, Fangyu Kang, Qingpeng Nong, Kaijin Zhang, Chaoyue Niu, Fan WuAbstract
Spatio-Temporal Video Grounding requires jointly localizing target objects across both temporal and spatial dimensions based on natural language queries, posing fundamental challenges for existing Multimodal Large Language Models (MLLMs). We identify two core challenges: \textit{entangled spatio-temporal alignment}, arising from coupling two heterogeneous sub-tasks within the same autoregressive output space, and \textit{dual-domain visual token redundancy}, where target objects exhibit simultaneous temporal and spatial sparsity, rendering the overwhelming majority of visual tokens irrelevant to the grounding query. To address these, we propose \textbf{Bridge-STG}, an end-to-end framework that decouples temporal and spatial localization while maintaining semantic coherence. While decoupling is the natural solution to this entanglement, it risks creating a semantic gap between the temporal MLLM and the spatial decoder. Bridge-STG resolves this through two pivotal designs: the \textbf{Spatio-Temporal Semantic Bridging (STSB)} mechanism with Explicit Temporal Alignment (ETA) distills the MLLM's temporal reasoning context into enriched bridging queries as a robust semantic interface; and the \textbf{Query-Guided Spatial Localization (QGSL)} module leverages these queries to drive a purpose-built spatial decoder with multi-layer interactive queries and positive/negative frame sampling, jointly eliminating dual-domain visual token redundancy. Extensive experiments across multiple benchmarks demonstrate that Bridge-STG achieves state-of-the-art performance among MLLM-based methods. Bridge-STG improves average m\_vIoU from $26.4$ to $34.3$ on VidSTG and demonstrates strong cross-task transfer across various fine-grained video understanding tasks under a unified multi-task training regime.
Abstract (translated)
URL
https://arxiv.org/abs/2604.08014
PDF
