vid-TLDR: Training Free Token merging for Light-weight Video Transformer

• URL: http://arxiv.org/abs/2403.13347v2
• Date: Sat, 30 Mar 2024 09:45:38 GMT
• Title: vid-TLDR: Training Free Token merging for Light-weight Video Transformer
• Authors: Joonmyung Choi, Sanghyeok Lee, Jaewon Chu, Minhyuk Choi, Hyunwoo J. Kim,
• Abstract summary: Video Transformers suffer from heavy computational costs induced by the massive number of tokens across the entire video frames. We propose training free token merging for lightweight video Transformer (vid-TLDR) We introduce the saliency-aware token merging strategy by dropping the background tokens and sharpening the object scores.
• Score: 14.143681665368856
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Video Transformers have become the prevalent solution for various video downstream tasks with superior expressive power and flexibility. However, these video transformers suffer from heavy computational costs induced by the massive number of tokens across the entire video frames, which has been the major barrier to training the model. Further, the patches irrelevant to the main contents, e.g., backgrounds, degrade the generalization performance of models. To tackle these issues, we propose training free token merging for lightweight video Transformer (vid-TLDR) that aims to enhance the efficiency of video Transformers by merging the background tokens without additional training. For vid-TLDR, we introduce a novel approach to capture the salient regions in videos only with the attention map. Further, we introduce the saliency-aware token merging strategy by dropping the background tokens and sharpening the object scores. Our experiments show that vid-TLDR significantly mitigates the computational complexity of video Transformers while achieving competitive performance compared to the base model without vid-TLDR. Code is available at https://github.com/mlvlab/vid-TLDR.

Related papers

• SViTT: Temporal Learning of Sparse Video-Text Transformers [65.93031164906812]
  We propose SViTT, a sparse video-text architecture that performs multi-frame reasoning with significantly lower cost than naive transformers with dense attention. SViTT employs two forms of sparsity: edge sparsity that limits the query-key communications between tokens in self-attention, and sparsity that discards uninformative visual tokens.
  arXiv  Detail & Related papers  (2023-04-18T08:17:58Z)
• On the Surprising Effectiveness of Transformers in Low-Labeled Video Recognition [18.557920268145818]
  Video vision transformers have been shown to be competitive with convolution-based methods (CNNs) broadly across multiple vision tasks. Our work empirically explores the low data regime for video classification and discovers that, surprisingly, transformers perform extremely well in the low-labeled video setting. We even show that using just the labeled data, transformers significantly outperform complex semi-supervised CNN methods that leverage large-scale unlabeled data as well.
  arXiv  Detail & Related papers  (2022-09-15T17:12:30Z)
• Pix4Point: Image Pretrained Standard Transformers for 3D Point Cloud Understanding [62.502694656615496]
  We present Progressive Point Patch Embedding and present a new point cloud Transformer model namely PViT. PViT shares the same backbone as Transformer but is shown to be less hungry for data, enabling Transformer to achieve performance comparable to the state-of-the-art. We formulate a simple yet effective pipeline dubbed "Pix4Point" that allows harnessing Transformers pretrained in the image domain to enhance downstream point cloud understanding.
  arXiv  Detail & Related papers  (2022-08-25T17:59:29Z)
• Efficient Attention-free Video Shift Transformers [56.87581500474093]
  This paper tackles the problem of efficient video recognition. Video transformers have recently dominated the efficiency (top-1 accuracy vs FLOPs) spectrum. We extend our formulation in the video domain to construct Video Affine-Shift Transformer.
  arXiv  Detail & Related papers  (2022-08-23T17:48:29Z)
• 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)
• All in One: Exploring Unified Video-Language Pre-training [44.22059872694995]
  We introduce an end-to-end video-language model, namely textitall-in-one Transformer, that embeds raw video and textual signals into joint representations. The code and pretrained model have been released in https://github.com/showlab/all-in-one.
  arXiv  Detail & Related papers  (2022-03-14T17:06:30Z)
• Video Transformers: A Survey [42.314208650554264]
  We study the contributions and trends for adapting Transformers to model video data. Specifically, we delve into how videos are embedded and tokenized, finding a very widspread use of large CNN backbones. Also, we analyse the self-supervised losses used to train Video Transformers, which to date are mostly constrained to contrastive approaches.
  arXiv  Detail & Related papers  (2022-01-16T07:31:55Z)
• VIOLET : End-to-End Video-Language Transformers with Masked Visual-token Modeling [88.30109041658618]
  A great challenge in video-language (VidL) modeling lies in the disconnection between fixed video representations extracted from image/video understanding models and downstream VidL data. We present VIOLET, a fully end-to-end VIdeO-LanguagE Transformer, which adopts a video transformer to explicitly model the temporal dynamics of video inputs.
  arXiv  Detail & Related papers  (2021-11-24T18:31:20Z)
• Spatiotemporal Transformer for Video-based Person Re-identification [102.58619642363958]
  We show that, despite the strong learning ability, the vanilla Transformer suffers from an increased risk of over-fitting. We propose a novel pipeline where the model is pre-trained on a set of synthesized video data and then transferred to the downstream domains. The derived algorithm achieves significant accuracy gain on three popular video-based person re-identification benchmarks.
  arXiv  Detail & Related papers  (2021-03-30T16:19:27Z)

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.