English
Related papers

Related papers: DUET-VLM: Dual stage Unified Efficient Token reduc…

200 papers

Multimodal Large Language Models (MLLMs) suffer from severe training inefficiency issue, which is associated with their massive model sizes and visual token numbers. Existing efforts in efficient training focus on reducing model sizes or…

Computer Vision and Pattern Recognition · Computer Science 2026-02-04 Dingkun Zhang , Shuhan Qi , Yulin Wu , Xinyu Xiao , Xuan Wang , Long Chen

Vision-Language Models (VLMs) process thousands of visual tokens per image alongside comparatively few text tokens, yet existing compression methods treat both modalities uniformly. We observe that the two modalities have fundamentally…

Machine Learning · Computer Science 2026-05-29 Yilin Feng , Ahmed Burak Gulhan , Mahmut Taylan Kandemir

Despite the impressive advancements of Large Vision-Language Models (LVLMs), existing approaches suffer from a fundamental bottleneck: inefficient visual-language integration. Current methods either disrupt the model's inherent structure or…

Computer Vision and Pattern Recognition · Computer Science 2025-06-23 Tongtian Yue , Longteng Guo , Yepeng Tang , Zijia Zhao , Xinxin Zhu , Hua Huang , Jing Liu

Pre-trained vision-language models (VLMs) have achieved impressive results in a range of vision-language tasks. However, popular VLMs usually consist of hundreds of millions of parameters which brings challenges for fine-tuning and…

Computation and Language · Computer Science 2022-10-17 Tiannan Wang , Wangchunshu Zhou , Yan Zeng , Xinsong Zhang

Video large language models (VideoLLM) excel at video understanding, but face efficiency challenges due to the quadratic complexity of abundant visual tokens. Our systematic analysis of token compression methods for VideoLLMs reveals two…

Computer Vision and Pattern Recognition · Computer Science 2025-11-19 Xuyang Liu , Yiyu Wang , Junpeng Ma , Linfeng Zhang

Large vision-language models (VLMs) typically process hundreds or thousands of visual tokens per image or video frame, incurring quadratic attention cost and substantial redundancy. Existing token reduction methods often ignore the textual…

Computer Vision and Pattern Recognition · Computer Science 2025-12-24 Kaitong Cai , Jusheng Zhang , Jing Yang , Yijia Fan , Pengtao Xie , Jian Wang , Keze Wang

Recent advancements in multimodal fusion have witnessed the remarkable success of vision-language (VL) models, which excel in various multimodal applications such as image captioning and visual question answering. However, building VL…

Computer Vision and Pattern Recognition · Computer Science 2024-10-24 Zhiwei Hao , Jianyuan Guo , Li Shen , Yong Luo , Han Hu , Yonggang Wen

Recent advancements in vision-language models (VLMs) have improved performance by increasing the number of visual tokens, which are often significantly longer than text tokens. However, we observe that most real-world scenarios do not…

Computer Vision and Pattern Recognition · Computer Science 2025-07-18 Senqiao Yang , Junyi Li , Xin Lai , Bei Yu , Hengshuang Zhao , Jiaya Jia

Large Vision-Language Models (LVLMs) excel in visual understanding and reasoning, but the excessive visual tokens lead to high inference costs. Although recent token reduction methods mitigate this issue, they mainly target single-turn…

Computer Vision and Pattern Recognition · Computer Science 2026-03-24 Yi Wang , Haofei Zhang , Qihan Huang , Anda Cao , Gongfan Fang , Wei Wang , Xuan Jin , Jie Song , Mingli Song , Xinchao Wang

The computational expense of redundant vision tokens in Large Vision-Language Models (LVLMs) has led many existing methods to compress them via a vision projector. However, this compression may lose visual information that is crucial for…

Computer Vision and Pattern Recognition · Computer Science 2025-11-27 Ze Feng , Jiang-jiang Liu , Sen Yang , Lingyu Xiao , Zhibin Quan , Zhenhua Feng , Wankou Yang , Jingdong Wang

In recent years, large-scale vision-language models (VLMs) have demonstrated remarkable performance on multimodal understanding and reasoning tasks. However, handling high-dimensional visual features often incurs substantial computational…

Computer Vision and Pattern Recognition · Computer Science 2025-12-23 Xiaoyang Guo , Keze Wang

In this paper, we introduce LightVLM, a simple but effective method that can be seamlessly deployed upon existing Vision-Language Models (VLMs) to greatly accelerate the inference process in a training-free manner. We divide the inference…

Computer Vision and Pattern Recognition · Computer Science 2025-09-03 Lianyu Hu , Fanhua Shang , Wei Feng , Liang Wan

Video large language models (VLLMs) have significantly advanced recently in processing complex video content, yet their inference efficiency remains constrained because of the high computational cost stemming from the thousands of visual…

Computer Vision and Pattern Recognition · Computer Science 2025-03-31 Keda Tao , Can Qin , Haoxuan You , Yang Sui , Huan Wang

Vision-Language Models (VLMs) excel in diverse visual tasks but face challenges in document understanding, which requires fine-grained text processing. While typical visual tasks perform well with low-resolution inputs, reading-intensive…

Computer Vision and Pattern Recognition · Computer Science 2024-12-13 Mor Shpigel Nacson , Aviad Aberdam , Roy Ganz , Elad Ben Avraham , Alona Golts , Yair Kittenplon , Shai Mazor , Ron Litman

Vision-Language Models (VLMs) are expensive because the LLM processes hundreds of largely redundant visual tokens. Existing token reduction methods typically exploit \textit{either} vision-encoder saliency (broad but query-agnostic)…

Computer Vision and Pattern Recognition · Computer Science 2026-02-03 Dhruv Parikh , Haoyang Fan , Rajgopal Kannan , Viktor Prasanna

A well-known dilemma in large vision-language models (e.g., GPT-4, LLaVA) is that while increasing the number of vision tokens generally enhances visual understanding, it also significantly raises memory and computational costs, especially…

Computer Vision and Pattern Recognition · Computer Science 2024-08-30 Shiwei Wu , Joya Chen , Kevin Qinghong Lin , Qimeng Wang , Yan Gao , Qianli Xu , Tong Xu , Yao Hu , Enhong Chen , Mike Zheng Shou

Large Multimodal Models (LMMs) have shown significant visual reasoning capabilities by connecting a visual encoder and a large language model. LMMs typically take in a fixed and large amount of visual tokens, such as the penultimate layer…

Computer Vision and Pattern Recognition · Computer Science 2026-02-03 Yuzhang Shang , Mu Cai , Bingxin Xu , Yong Jae Lee , Yan Yan

Existing Multimodal Large Language Models (MLLMs) process a large number of visual tokens, leading to significant computational costs and inefficiency. Instruction-related visual token compression demonstrates strong task relevance, which…

Computer Vision and Pattern Recognition · Computer Science 2026-05-05 Lei Lei , Jie Gu , Xiaokang Ma , Chu Tang , Jingmin Chen , Tong Xu

We present DyMU, an efficient, training-free framework that dynamically reduces the computational burden of vision-language models (VLMs) while maintaining high task performance. Our approach comprises two key components. First, Dynamic…

Computer Vision and Pattern Recognition · Computer Science 2025-05-13 Zhenhailong Wang , Senthil Purushwalkam , Caiming Xiong , Silvio Savarese , Heng Ji , Ran Xu

Video large language models (Video-LLMs) face high computational costs due to large volumes of visual tokens. Existing token compression methods typically adopt a two-stage spatiotemporal compression strategy, relying on stage-specific…

Computer Vision and Pattern Recognition · Computer Science 2026-03-24 Junhao Du , Jialong Xue , Anqi Li , Jincheng Dai , Guo Lu