Beyond Training: Dynamic Token Merging for Zero-Shot Video Understanding

• URL: http://arxiv.org/abs/2411.14401v1
• Date: Thu, 21 Nov 2024 18:30:11 GMT
• Title: Beyond Training: Dynamic Token Merging for Zero-Shot Video Understanding
• Authors: Yiming Zhang, Zhuokai Zhao, Zhaorun Chen, Zenghui Ding, Xianjun Yang, Yining Sun,
• Abstract summary: We propose DYTO, a novel dynamic token merging framework for zero-shot video understanding. DYTO integrates a hierarchical frame selection and a bipartite token merging strategy to dynamically cluster key frames and selectively compress token sequences. Experiments demonstrate the effectiveness of DYTO, achieving superior performance compared to both fine-tuned and training-free methods.
• Score: 11.211803499867639
• License:
• Abstract: Recent advancements in multimodal large language models (MLLMs) have opened new avenues for video understanding. However, achieving high fidelity in zero-shot video tasks remains challenging. Traditional video processing methods rely heavily on fine-tuning to capture nuanced spatial-temporal details, which incurs significant data and computation costs. In contrast, training-free approaches, though efficient, often lack robustness in preserving context-rich features across complex video content. To this end, we propose DYTO, a novel dynamic token merging framework for zero-shot video understanding that adaptively optimizes token efficiency while preserving crucial scene details. DYTO integrates a hierarchical frame selection and a bipartite token merging strategy to dynamically cluster key frames and selectively compress token sequences, striking a balance between computational efficiency with semantic richness. Extensive experiments across multiple benchmarks demonstrate the effectiveness of DYTO, achieving superior performance compared to both fine-tuned and training-free methods and setting a new state-of-the-art for zero-shot video understanding.

Related papers

• DynFocus: Dynamic Cooperative Network Empowers LLMs with Video Understanding [18.312501339046296]
  We observe that redundancy occurs in both repeated and answer-irrelevant frames, and the corresponding frames vary with different questions. This suggests the possibility of adopting dynamic encoding to balance detailed video information preservation with token budget reduction.
  arXiv  Detail & Related papers  (2024-11-19T09:16:54Z)
• Free Video-LLM: Prompt-guided Visual Perception for Efficient Training-free Video LLMs [56.040198387038025]
  We present a novel prompt-guided visual perception framework (abbreviated as Free Video-LLM) for efficient inference of training-free video LLMs. Our method effectively reduces the number of visual tokens while maintaining high performance across multiple video question-answering benchmarks.
  arXiv  Detail & Related papers  (2024-10-14T12:35:12Z)
• Realizing Video Summarization from the Path of Language-based Semantic Understanding [19.825666473712197]
  We propose a novel video summarization framework inspired by the Mixture of Experts (MoE) paradigm. Our approach integrates multiple VideoLLMs to generate comprehensive and coherent textual summaries.
  arXiv  Detail & Related papers  (2024-10-06T15:03:22Z)
• Video-LaVIT: Unified Video-Language Pre-training with Decoupled Visual-Motional Tokenization [52.63845811751936]
  Video pre-training is challenging due to the modeling of its dynamics video. In this paper, we address such limitations in video pre-training with an efficient video decomposition. Our framework is both capable of comprehending and generating image and video content, as demonstrated by its performance across 13 multimodal benchmarks.
  arXiv  Detail & Related papers  (2024-02-05T16:30:49Z)
• Towards Efficient and Effective Text-to-Video Retrieval with Coarse-to-Fine Visual Representation Learning [15.998149438353133]
  We propose a two-stage retrieval architecture for text-to-video retrieval. In training phase, we design a parameter-free text-gated interaction block (TIB) for fine-grained video representation learning. In retrieval phase, we use coarse-grained video representations for fast recall of top-k candidates, which are then reranked by fine-grained video representations.
  arXiv  Detail & Related papers  (2024-01-01T08:54:18Z)
• Leaping Into Memories: Space-Time Deep Feature Synthesis [93.10032043225362]
  We propose LEAPS, an architecture-independent method for synthesizing videos from internal models. We quantitatively and qualitatively evaluate the applicability of LEAPS by inverting a range of architectures convolutional attention-based on Kinetics-400.
  arXiv  Detail & Related papers  (2023-03-17T12:55:22Z)
• Deep Unsupervised Key Frame Extraction for Efficient Video Classification [63.25852915237032]
  This work presents an unsupervised method to retrieve the key frames, which combines Convolutional Neural Network (CNN) and Temporal Segment Density Peaks Clustering (TSDPC) The proposed TSDPC is a generic and powerful framework and it has two advantages compared with previous works, one is that it can calculate the number of key frames automatically. Furthermore, a Long Short-Term Memory network (LSTM) is added on the top of the CNN to further elevate the performance of classification.
  arXiv  Detail & Related papers  (2022-11-12T20:45:35Z)
• 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)
• Contrastive Transformation for Self-supervised Correspondence Learning [120.62547360463923]
  We study the self-supervised learning of visual correspondence using unlabeled videos in the wild. Our method simultaneously considers intra- and inter-video representation associations for reliable correspondence estimation. Our framework outperforms the recent self-supervised correspondence methods on a range of visual tasks.
  arXiv  Detail & Related papers  (2020-12-09T14:05:06Z)
• Temporal Context Aggregation for Video Retrieval with Contrastive Learning [81.12514007044456]
  We propose TCA, a video representation learning framework that incorporates long-range temporal information between frame-level features. The proposed method shows a significant performance advantage (17% mAP on FIVR-200K) over state-of-the-art methods with video-level features.
  arXiv  Detail & Related papers  (2020-08-04T05:24:20Z)

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.