Linear Scaling Video VLMs for Long Video Understanding

• URL: http://arxiv.org/abs/2605.31598v1
• Date: Fri, 29 May 2026 17:59:02 GMT
• Title: Linear Scaling Video VLMs for Long Video Understanding
• Abstract summary: Video encoders still rely on self-attention, causing compute and latency to grow quadratically with the number of frames.<n>We introduce StateKV, an inference-time method that adapts pretrained long-videoMs to linear-time video prefill.
• Score: 26.67642601439313
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Video vision-language models (VLMs) are increasingly used in long-horizon and streaming settings, yet most video encoders still rely on spatiotemporal self-attention, causing compute and latency to grow quadratically with the number of frames. Existing efficiency methods improve scalability but often lose accuracy relative to full self-attention, for example through aggressive frame/token dropping or coarse attention approximations. We introduce StateKV, an inference-time method that adapts pretrained long-video VLMs to linear-time video prefill by carrying cross-frame context in a fixed-capacity, importance-based recurrent state, paired with a second full per-frame cache used for decoding. Across three long-video benchmarks and seven models spanning three families and multiple scales, StateKV remains close to full self-attention and consistently outperforms dominant sliding-window / recency-based streaming approximations, without fine-tuning or architectural changes. StateKV also reduces video-prefill cost measured FLOPs, enabling stronger accuracy at a fixed compute budget by running larger models. These results suggest a practical step toward scalable long-video understanding.

Related papers

• LongE2V: Long-Horizon Event-based Video Reconstruction, Prediction, and Frame Interpolation with Video Diffusion Models [14.33633321388011]
  LongE2V is a novel approach that leverages pre-trained video diffusion priors to handle event-based video reconstruction.<n>By fine-tuning agressive video models, our approach mitigates high data efficiency and superior perceptual quality.<n>Experiments on real-world benchmarks demonstrate that LongE2V outperforms state-of-the-art methods across all three tasks.
  arXiv  Detail & Related papers  (2026-07-09T17:59:52Z)
• Attend Locally, Remember Linearly: Linear Attention as Cross-Frame Memory for Autoregressive Video Diffusion [61.57938553036056]
  ARL2 is a hybrid attention module that replaces quadratic cross-frame attention with a fixed-size recurrent state.<n>We decompose self-attention into two branches: an intra-frame softmax branch for spatial detail and local dependencies, and an inter-frame recurrent linear branch that maintains a fixed-size state for streaming context.<n>With 75% of layers replaced by hybrid linear attention, the model achieves up to 2.26 wall-clock speedup and 54% memory reduction, while maintaining comparable quality with improving temporal consistency.
  arXiv  Detail & Related papers  (2026-05-15T19:33:45Z)
• Towards Holistic Modeling for Video Frame Interpolation with Auto-regressive Diffusion Transformers [95.68243351895107]
  We propose a holistic, video-centric paradigm named textbfLocal textbfDiffusion textbfForcing for textbfVideo textbfFrame textbfInterpolation (LDF-VFI)<n>Our framework is built upon an auto-regressive diffusion transformer that models the entire video sequence to ensure long-range temporal coherence.<n>LDF-VFI achieves state-of-the-art performance on challenging long-sequence benchmarks, demonstrating superior per
  arXiv  Detail & Related papers  (2026-01-21T12:58:52Z)
• Efficient Video Sampling: Pruning Temporally Redundant Tokens for Faster VLM Inference [5.146388234814547]
  Long videos often exceed the token budget of modern language models, leading to severe context limitations and latency issues.<n>We introduce Efficient Video Sampling (EVS), a simple, plug-and-play method for reducing token redundancy in videos by identifying and pruning temporally static patches.<n>EVS substantially reduces token count while maintaining semantic fidelity, enabling faster inference and longer input sequences.
  arXiv  Detail & Related papers  (2025-10-16T12:34:38Z)
• Exploiting Temporal State Space Sharing for Video Semantic Segmentation [53.8810901249897]
  Video semantic segmentation (VSS) plays a vital role in understanding the temporal evolution of scenes.<n>Traditional methods often segment videos frame-by-frame or in a short temporal window, leading to limited temporal context, redundant computations, and heavy memory requirements.<n>We introduce a Temporal Video State Space Sharing architecture to leverage Mamba state space models for temporal feature sharing.<n>Our model features a selective gating mechanism that efficiently propagates relevant information across video frames, eliminating the need for a memory-heavy feature pool.
  arXiv  Detail & Related papers  (2025-03-26T01:47:42Z)
• STORM: Token-Efficient Long Video Understanding for Multimodal LLMs [116.4479155699528]
  STORM is a novel architecture incorporating a dedicated temporal encoder between the image encoder and the Video-LLMs.<n>We show that STORM achieves state-of-the-art results across various long video understanding benchmarks.
  arXiv  Detail & Related papers  (2025-03-06T06:17:38Z)
• Making Every Frame Matter: Continuous Activity Recognition in Streaming Video via Adaptive Video Context Modeling [19.205142489726875]
  Video activity recognition has become increasingly important in robots and embodied AI.<n>We introduce a novel system, CARS, to overcome these issues through adaptive video context modeling.<n>Our CARS runs at speeds $>$30 FPS on typical edge devices and outperforms all baselines by 1.2% to 79.7% in accuracy.
  arXiv  Detail & Related papers  (2024-10-19T05:50:00Z)
• A Simple Recipe for Contrastively Pre-training Video-First Encoders Beyond 16 Frames [57.758863967770594]
  We build on the common paradigm of transferring large-scale, image--text models to video via shallow temporal fusion.<n>We expose two limitations to the approach: (1) decreased spatial capabilities, likely due to poor video--language alignment in standard video datasets, and (2) higher memory consumption, bottlenecking the number of frames that can be processed.
  arXiv  Detail & Related papers  (2023-12-12T16:10:19Z)
• Learning Trajectory-Aware Transformer for Video Super-Resolution [50.49396123016185]
  Video super-resolution aims to restore a sequence of high-resolution (HR) frames from their low-resolution (LR) counterparts. Existing approaches usually align and aggregate video frames from limited adjacent frames. We propose a novel Transformer for Video Super-Resolution (TTVSR)
  arXiv  Detail & Related papers  (2022-04-08T03:37:39Z)

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.