{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T09:57:43Z","timestamp":1777888663155,"version":"3.51.4"},"reference-count":66,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,10,19]],"date-time":"2025-10-19T00:00:00Z","timestamp":1760832000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,10,19]],"date-time":"2025-10-19T00:00:00Z","timestamp":1760832000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62422204,62125201,U24B20174,62402152,62406093"],"award-info":[{"award-number":["62422204,62125201,U24B20174,62402152,62406093"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100022963","name":"Key Research and Development Program of Zhejiang Province","doi-asserted-by":"publisher","award":["2025C01026"],"award-info":[{"award-number":["2025C01026"]}],"id":[{"id":"10.13039\/100022963","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100004731","name":"Zhejiang Provincial Natural Science Foundation of China","doi-asserted-by":"publisher","award":["LDT23F02025F02,LQN25F020017,LQ24F020032"],"award-info":[{"award-number":["LDT23F02025F02,LQN25F020017,LQ24F020032"]}],"id":[{"id":"10.13039\/501100004731","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,10,19]]},"DOI":"10.1109\/iccv51701.2025.01866","type":"proceedings-article","created":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T19:45:49Z","timestamp":1777491949000},"page":"20064-20074","source":"Crossref","is-referenced-by-count":0,"title":["Growing a Twig to Accelerate Large Vision-Language Models"],"prefix":"10.1109","author":[{"given":"Zhenwei","family":"Shao","sequence":"first","affiliation":[{"name":"School of Computer Science, Hangzhou Dianzi University,Zhejiang Key Laboratory of Space Information Sensing and Transmission,Hangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Mingyang","family":"Wang","sequence":"additional","affiliation":[{"name":"School of Computer Science, Hangzhou Dianzi University,Zhejiang Key Laboratory of Space Information Sensing and Transmission,Hangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zhou","family":"Yu","sequence":"additional","affiliation":[{"name":"School of Computer Science, Hangzhou Dianzi University,Zhejiang Key Laboratory of Space Information Sensing and Transmission,Hangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wenwen","family":"Pan","sequence":"additional","affiliation":[{"name":"School of Computer Science, Hangzhou Dianzi University,Zhejiang Key Laboratory of Space Information Sensing and Transmission,Hangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yan","family":"Yang","sequence":"additional","affiliation":[{"name":"School of Computer Science, Hangzhou Dianzi University,Zhejiang Key Laboratory of Space Information Sensing and Transmission,Hangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Tao","family":"Wei","sequence":"additional","affiliation":[{"name":"Li Auto Inc.,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Hongyuan","family":"Zhang","sequence":"additional","affiliation":[{"name":"Li Auto Inc.,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Ning","family":"Mao","sequence":"additional","affiliation":[{"name":"Li Auto Inc.,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wei","family":"Chen","sequence":"additional","affiliation":[{"name":"Li Auto Inc.,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Jun","family":"Yu","sequence":"additional","affiliation":[{"name":"School of Computer Science, Hangzhou Dianzi University,Zhejiang Key Laboratory of Space Information Sensing and Transmission,Hangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Qwen-vl: A frontier large vision-language model with versatile abilities","author":"Bai","year":"2023","journal-title":"arXiv preprint"},{"key":"ref2","article-title":"Token merging: Your vit but faster","volume-title":"The Eleventh International Conference on Learning Representations","author":"Bolya","year":"2023"},{"key":"ref3","article-title":"Rt-2: Vision-language-action models transfer web knowledge to robotic control","author":"Brohan","year":"2023","journal-title":"arXiv preprint"},{"key":"ref4","first-page":"1877","article-title":"Language models are few-shot learners","volume-title":"Advances in Neural Information Processing Systems","author":"Brown","year":"2020"},{"key":"ref5","article-title":"Medusa: Simple llm inference acceleration framework with multiple decoding heads","author":"Cai","year":"2024","journal-title":"arXiv preprint"},{"key":"ref6","article-title":"Auroracap: Efficient, performant video detailed captioning and a new benchmark","author":"Chai","year":"2024","journal-title":"arXiv preprint"},{"key":"ref7","article-title":"Llavolta: Efficient multi-modal models via stage-wise visual context compression","author":"Chen","year":"2024","journal-title":"arXiv preprint"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-73004-7_2"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1007\/s11432-024-4231-5"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/DAC56929.2023.10247993"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1189"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.acl-long.681"},{"key":"ref13","article-title":"Mme: A comprehensive evaluation benchmark for multimodal large language models","author":"Fu","year":"2023","journal-title":"arXiv preprint"},{"key":"ref14","first-page":"8285","article-title":"On speculative decoding for multimodal large language models","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Gagrani","year":"2024"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.670"},{"key":"ref16","article-title":"Mamba: Linear-time sequence modeling with selective state spaces","volume-title":"First Conference on Language Modeling","author":"Gu","year":"2024"},{"key":"ref17","article-title":"Deepseek-r1: Incentivizing reasoning capability in 11 ms via reinforcement learning","author":"Guo","year":"2025","journal-title":"arXiv preprint"},{"key":"ref18","article-title":"Rethinking token reduction in mllms: Towards a unified paradigm for training-free acceleration","author":"Han","year":"2024","journal-title":"arXiv preprint"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00686"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.149"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.emnlp-main.825"},{"key":"ref22","article-title":"Openvla: An open-source vision-language-action model","author":"Kim","year":"2024","journal-title":"arXiv preprint"},{"key":"ref23","first-page":"19274","article-title":"Fast inference from transformers via speculative decoding","volume-title":"International Conference on Machine Learning","author":"Leviathan","year":"2023"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/tpami.2025.3637265"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.emnlp-main.342"},{"key":"ref26","article-title":"Deepseek-v3 technical report","author":"Liu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref27","first-page":"11946","article-title":"Kangaroo: Lossless self-speculative decoding for accelerating 11 ms via double early exiting","volume":"37","author":"Liu","year":"2025","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.52202\/075280-1516"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02484"},{"key":"ref30","author":"Liu","year":"2024","journal-title":"Llava-next: Improved reasoning, ocr, and world knowledge"},{"key":"ref31","article-title":"Multi-stage vision token dropping: Towards efficient multimodal large language model","author":"Liu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-72658-3_13"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.52202\/068431-0182"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.01480"},{"key":"ref35","first-page":"417","article-title":"Groma: Localized visual tokenization for grounding multimodal large language models","volume-title":"European Conference on Computer Vision","author":"Ma","year":"2024"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.52202\/075280-0848"},{"key":"ref37","volume-title":"Gpt-4v(ision) system card","year":"2023"},{"key":"ref38","article-title":"Grounding multimodal large language models to the world","volume-title":"The Twelfth International Conference on Learning Representations","author":"Peng","year":"2024"},{"key":"ref39","article-title":"Llava-prumerge: Adaptive token reduction for efficient large multimodal models","author":"Shang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1109\/TMM.2025.3557680"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00851"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.52202\/079017-0235"},{"key":"ref43","article-title":"Llama 2: Open foundation and fine-tuned chat models","author":"Touvron","year":"2023","journal-title":"arXiv preprint"},{"key":"ref44","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52734.2025.01841"},{"key":"ref45","article-title":"Attention is all you need","volume-title":"Advances in Neural Information Processing Systems","author":"Vaswani","year":"2017"},{"key":"ref46","article-title":"Mobile-agent: Autonomous multi-modal mobile device agent with visual perception","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref47","article-title":"Accelerating multimodal large language models via dynamic visual-token exit and the empirical findings","author":"Wu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref48","article-title":"Efficient streaming language models with attention sinks","volume-title":"The Twelfth International Conference on Learning Representations","author":"Xiao","year":"2024"},{"key":"ref49","article-title":"Pyramiddrop: Accelerating your large vision-language models via pyramid visual redundancy reduction","author":"Xing","year":"2024","journal-title":"arXiv preprint"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1145\/3123266.3123427"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52734.2025.01843"},{"key":"ref52","article-title":"Minicpm-v: A gpt-4v level mllm on your phone","author":"Yao","year":"2024","journal-title":"arXiv preprint"},{"key":"ref53","article-title":"mplug-docowl: Modularized multimodal large language model for document understanding","author":"Ye","year":"2023","journal-title":"arXiv preprint"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v39i21.34366"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52734.2025.02325"},{"key":"ref56","article-title":"Mm-vet: Evaluating large multimodal models for integrated capabilities","author":"Yu","year":"2023","journal-title":"arXiv preprint"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v33i01.33019127"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1145\/3706598.3713600"},{"key":"ref59","article-title":"[cls] attention is all you need for trainingfree visual token pruning: Make vlm inference faster","author":"Zhang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref60","article-title":"Sparsevlm: Visual token sparsification for efficient vision-language model inference","author":"Zhang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref61","doi-asserted-by":"publisher","DOI":"10.52202\/075280-1506"},{"key":"ref62","article-title":"Treat visual tokens as text? but your mllm only needs fewer efforts to see","author":"Zhang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref63","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52734.2025.01842"},{"key":"ref64","article-title":"A stitch in time saves nine: Small vlm is a precise guidance for accelerating large vlms","author":"Zhao","year":"2024","journal-title":"arXiv preprint"},{"key":"ref65","article-title":"A survey on efficient inference for large language models","author":"Zhou","year":"2024","journal-title":"arXiv preprint"},{"key":"ref66","article-title":"MiniGPT-4: Enhancing vision-language understanding with advanced large language models","volume-title":"The Twelfth International Conference on Learning Representations","author":"Zhu","year":"2024"}],"event":{"name":"2025 IEEE\/CVF International Conference on Computer Vision (ICCV)","location":"Honolulu, HI, USA","start":{"date-parts":[[2025,10,19]]},"end":{"date-parts":[[2025,10,25]]}},"container-title":["2025 IEEE\/CVF International Conference on Computer Vision (ICCV)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11443115\/11443287\/11444911.pdf?arnumber=11444911","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T04:54:38Z","timestamp":1777611278000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11444911\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,19]]},"references-count":66,"URL":"https:\/\/doi.org\/10.1109\/iccv51701.2025.01866","relation":{},"subject":[],"published":{"date-parts":[[2025,10,19]]}}}