ReTaKe: Reducing Temporal and Knowledge Redundancy for Long Video Understanding

• URL: http://arxiv.org/abs/2412.20504v4
• Date: Sun, 16 Mar 2025 16:25:31 GMT
• Title: ReTaKe: Reducing Temporal and Knowledge Redundancy for Long Video Understanding
• Authors: Xiao Wang, Qingyi Si, Jianlong Wu, Shiyu Zhu, Li Cao, Liqiang Nie,
• Abstract summary: ReTaKe enables VideoLLMs to process 8 times longer frames (up to 2048), similar-sized models by 3-5% and even rivaling much larger ones on VideoMME, MLVU, LongVideoBench, and LVBench.<n>Our code is available at https://github.com/SCZwangxiao/video-ReTaKe.
• Score: 55.320254859515714
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Video Large Language Models (VideoLLMs) have made significant strides in video understanding but struggle with long videos due to the limitations of their backbone LLMs. Existing solutions rely on length extrapolation, which is memory-constrained, or visual token compression, which primarily leverages low-level temporal redundancy while overlooking the more effective high-level knowledge redundancy. To address this, we propose $\textbf{ReTaKe}$, a training-free method with two novel modules DPSelect and PivotKV, to jointly reduce both temporal visual redundancy and knowledge redundancy for video compression. To align with the way of human temporal perception, DPSelect identifies keyframes based on inter-frame distance peaks. To leverage LLMs' learned prior knowledge, PivotKV marks the keyframes as pivots and compress non-pivot frames by pruning low-attention tokens in their KV cache. ReTaKe enables VideoLLMs to process 8 times longer frames (up to 2048), outperforming similar-sized models by 3-5% and even rivaling much larger ones on VideoMME, MLVU, LongVideoBench, and LVBench. Moreover, by overlapping compression operations with prefilling, ReTaKe introduces only ~10% prefilling latency overhead while reducing decoding latency by ~20%. Our code is available at https://github.com/SCZwangxiao/video-ReTaKe.

Related papers

• AdaReTaKe: Adaptive Redundancy Reduction to Perceive Longer for Video-language Understanding [55.320254859515714]
  Multimodal Large Language Models (MLLMs) have revolutionized video understanding, yet are still limited by context length when processing long videos. We propose AdaReTaKe, a training-free method that flexibly reduces visual redundancy by allocating compression ratios among time and layers with theoretical guarantees. Experiments on VideoMME, MLVU, LongVideoBench, and LVBench datasets demonstrate that AdaReTaKe outperforms existing methods by 2.3% and 2.8% for 7B and 72B models, respectively.
  arXiv  Detail & Related papers  (2025-03-16T16:14:52Z)
• BIMBA: Selective-Scan Compression for Long-Range Video Question Answering [46.199493246921435]
  Video Question Answering (VQA) in long videos poses the key challenge of extracting relevant information. We introduce BIMBA, an efficient state-space model to handle long-form videos.
  arXiv  Detail & Related papers  (2025-03-12T17:57:32Z)
• Token-Efficient Long Video Understanding for Multimodal LLMs [101.70681093383365]
  STORM is a novel architecture incorporating a dedicated temporal encoder between the image encoder and the Video-LLMs. 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)
• AdaCM$^2$: On Understanding Extremely Long-Term Video with Adaptive Cross-Modality Memory Reduction [10.579335027350263]
  AdaCM$2$ is an adaptive cross-modality memory reduction approach to video-text alignment on video streams. It achieves a 4.5% improvement across multiple tasks in the LVU dataset with a GPU memory consumption reduction of up to 65%.
  arXiv  Detail & Related papers  (2024-11-19T18:04:13Z)
• LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding [65.46303012350207]
  LongVU is an adaptive compression mechanism that reduces the number of video tokens while preserving visual details of long videos. We leverage DINOv2 features to remove redundant frames that exhibit high similarity. We perform spatial token reduction across frames based on their temporal dependencies.
  arXiv  Detail & Related papers  (2024-10-22T21:21:37Z)
• Video-XL: Extra-Long Vision Language Model for Hour-Scale Video Understanding [25.61734041983714]
  Video-XL is a novel approach that leverages MLLMs' inherent key-value sparsification capacity to condense the visual input.<n>Video-XL's effectiveness is verified from three aspects. First, it achieves a superior long-video understanding capability, outperforming state-of-the-art models of comparable sizes.
  arXiv  Detail & Related papers  (2024-09-22T15:13:31Z)
• Koala: Key frame-conditioned long video-LLM [70.52369588364992]
  We propose a lightweight and self-supervised long video-LLM (Koala) to adapt pretrained vLLMs for generalizing to longer videos. Our approach outperforms state-of-the-art large models by 3 - 6% in absolute accuracy across all tasks. Surprisingly, we also empirically show that our approach not only helps a pretrained vLLM to understand long videos but also improves its accuracy on short-term action recognition.
  arXiv  Detail & Related papers  (2024-04-05T18:33:04Z)
• 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)
• Use Your Head: Improving Long-Tail Video Recognition [28.506807977493434]
  We demonstrate that, unlike naturally-collected video datasets and existing long-tail image benchmarks, current video benchmarks fall short on multiple long-tailed properties. We propose new video benchmarks that better assess long-tail recognition, by sampling subsets from two datasets: SSv2 and VideoLT.
  arXiv  Detail & Related papers  (2023-04-03T17:09:47Z)
• Frame-wise Action Representations for Long Videos via Sequence Contrastive Learning [44.412145665354736]
  We introduce a novel contrastive action representation learning framework to learn frame-wise action representations. Inspired by the recent progress of self-supervised learning, we present a novel sequence contrastive loss (SCL) applied on two correlated views. Our approach also shows outstanding performance on video alignment and fine-grained frame retrieval tasks.
  arXiv  Detail & Related papers  (2022-03-28T17:59:54Z)
• Learning for Video Compression with Recurrent Auto-Encoder and Recurrent Probability Model [164.7489982837475]
  This paper proposes a Recurrent Learned Video Compression (RLVC) approach with the Recurrent Auto-Encoder (RAE) and Recurrent Probability Model ( RPM) The RAE employs recurrent cells in both the encoder and decoder to exploit the temporal correlation among video frames. Our approach achieves the state-of-the-art learned video compression performance in terms of both PSNR and MS-SSIM.
  arXiv  Detail & Related papers  (2020-06-24T08:46:33Z)

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.