{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T10:02:36Z","timestamp":1777888956588,"version":"3.51.4"},"reference-count":80,"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"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,10,19]]},"DOI":"10.1109\/iccv51701.2025.02238","type":"proceedings-article","created":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T19:45:49Z","timestamp":1777491949000},"page":"24147-24158","source":"Crossref","is-referenced-by-count":0,"title":["Token-Efficient VLM: High-Resolution Image Understanding Via Dynamic Region Proposal"],"prefix":"10.1109","author":[{"given":"Yitong","family":"Jiang","sequence":"first","affiliation":[{"name":"NVIDIA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Jinwei","family":"Gu","sequence":"additional","affiliation":[{"name":"NVIDIA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Tianfan","family":"Xue","sequence":"additional","affiliation":[{"name":"The Chinese University of Hong Kong"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Ka Chun","family":"Cheung","sequence":"additional","affiliation":[{"name":"NVIDIA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Pavlo","family":"Molchanov","sequence":"additional","affiliation":[{"name":"NVIDIA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Hongxu","family":"Yin","sequence":"additional","affiliation":[{"name":"NVIDIA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Sifei","family":"Liu","sequence":"additional","affiliation":[{"name":"NVIDIA"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1723"},{"key":"ref2","first-page":"1","article-title":"Qwen-vl: A frontier large vision-language model with versatile abilities","author":"Bai","year":"2023","journal-title":"arXiv preprint"},{"key":"ref3","first-page":"4","volume-title":"Coyo- 700 m: Image-text pair dataset","author":"Byeon","year":"2022"},{"key":"ref4","first-page":"5","article-title":"Sharegpt4v: Improving large multi-modal models with better captions","author":"Chen","year":"2023","journal-title":"arXiv preprint"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52733.2024.01368"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.52202\/079017-4293"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.52202\/075280-2142"},{"key":"ref8","article-title":"Nvlm: Open frontier-class multimodal 11 ms","author":"Dai","year":"2024","journal-title":"arXiv preprint"},{"key":"ref9","article-title":"Internlm-xcomposer24khd: A pioneering large vision-language model handling resolutions from 336 pixels to 4 k hd","author":"Dong","year":"2024","journal-title":"arXiv preprint"},{"key":"ref10","article-title":"An image is worth 16 \u00d7 16 words: Transformers for image recognition at scale","author":"Dosovitskiy","year":"2020","journal-title":"arXiv preprint"},{"key":"ref11","article-title":"vila2: Vila augmented vila","author":"Fang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref12","first-page":"1","article-title":"Mme: A\\\\ comprehensive evaluation benchmark for multimodal large language models","author":"Fu","year":"2023","journal-title":"ArXiv, abs\/2306.13394"},{"key":"ref13","article-title":"G-llava: Solving geometric problem with multi-modal large language model","author":"Gao","year":"2023","journal-title":"arXiv preprint"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.670"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.01309"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-20059-5_32"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00686"},{"key":"ref18","year":"2023","journal-title":"Introducing idefics: An open reproduction of state-of-the-art visual language model."},{"key":"ref19","article-title":"Spatial transformer networks","volume-title":"Proceedings of Advances in Neural Information Processing Systems (NeurIPS)","volume":"28","author":"Jaderberg"},{"key":"ref20","first-page":"2901","article-title":"Lawrence Zitnick, and Ross Girshick","volume-title":"Clevr: A diagnostic dataset for compositional language and elementary visual reasoning. In Proceedings of the IEEE conference on computer vision and pattern recognition","author":"Johnson"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00592"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-46493-0_15"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-19815-1_29"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.356"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1145\/3477495.3531753"},{"key":"ref26","first-page":"1","article-title":"Seed-bench: Benchmarking multimodal 11 ms with generative comprehension","author":"Li","year":"2023","journal-title":"arXiv preprint"},{"key":"ref27","article-title":"Llava-onevision: Easy visual task transfer","author":"Li","year":"2024","journal-title":"arXiv preprint"},{"key":"ref28","article-title":"Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models","author":"Li","year":"2023","journal-title":"In ICML"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.emnlp-main.20"},{"key":"ref30","first-page":"2","article-title":"Mini-gemini: Mining the potential of multi-modality vision language models","author":"Li","year":"2024","journal-title":"arXiv preprint"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02527"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02520"},{"key":"ref33","first-page":"1","article-title":"Sphinx: The joint mixing of weights, tasks, and visual embeddings for multi-modal large language models","author":"Lin","year":"2023","journal-title":"arXiv preprint"},{"key":"ref34","article-title":"Mmc: Advancing multimodal chart understanding with large-scale instruction tuning","author":"Liu","year":"2023","journal-title":"arXiv preprint"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02484"},{"key":"ref36","first-page":"1","author":"Liu","year":"2024","journal-title":"Llava-next: Improved reasoning, ocr, and world knowledge"},{"key":"ref37","first-page":"1","article-title":"Visual instruction tuning","author":"Liu","year":"2024","journal-title":"NIPS"},{"key":"ref38","first-page":"1","article-title":"Infimm-hd: A leap forward in high-resolution multimodal understanding","author":"Liu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref39","first-page":"3","article-title":"Grounding dino: Marrying dino with grounded pre-training for open-set object detection","author":"Liu","year":"2023","journal-title":"arXiv preprint"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1109\/tpami.2026.3653415"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-72658-3_13"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52734.2025.00390"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.52202\/068431-0182"},{"key":"ref44","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00331"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.findings-acl.177"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/WACV48630.2021.00225"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1109\/WACV51458.2022.00264"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1109\/ICDAR.2019.00156"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.3403\/00116074"},{"issue":"5","key":"ref50","first-page":"1","volume-title":"Gpt-4 technical report. arxiv 2303.08774. View in Article","volume":"2","year":"2023"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1833"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-73242-3_25"},{"key":"ref53","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-58536-5_44"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00851"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1145\/3404835.3463257"},{"key":"ref56","article-title":"Gemini: a family of highly capable multimodal models","author":"Team","year":"2023","journal-title":"arXiv preprint"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.52202\/079017-2771"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52733.2024.00914"},{"key":"ref59","article-title":"Llama: Open and efficient foundation language models","author":"Touvron","year":"2023","journal-title":"arXiv preprint"},{"key":"ref60","article-title":"Finetuned language models are zero-shot learners","author":"Wei","year":"2021","journal-title":"arXiv preprint"},{"key":"ref61","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52733.2024.01243"},{"key":"ref62","article-title":"Sheared llama: Accelerating language model pre-training via structured pruning","author":"Xia","year":"2023","journal-title":"arXiv preprint"},{"key":"ref63","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.00475"},{"key":"ref64","article-title":"Llava-uhd: an 1 mm perceiving any aspect ratio and high-resolution images","author":"Xu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref65","article-title":"Qwen2 technical report","volume-title":"arXiv preprint","author":"Yang","year":"2024"},{"key":"ref66","first-page":"2","article-title":"Ferret: Refer and ground anything anywhere at any granularity","author":"You","year":"2023","journal-title":"arXiv preprint"},{"key":"ref67","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV48922.2021.00044"},{"key":"ref68","article-title":"Mammoth: Building math generalist models through hybrid instruction tuning","author":"Yue","year":"2023","journal-title":"arXiv preprint"},{"key":"ref69","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.00913"},{"key":"ref70","article-title":"Halle-switch: Controlling object hallucination in large vision language models","author":"Zhai","year":"2023","journal-title":"arXiv e-prints"},{"key":"ref71","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.01100"},{"key":"ref72","first-page":"2","article-title":"Griffon-g: Bridging visionlanguage and vision-centric tasks via large multimodal models","author":"Zhan","year":"2024","journal-title":"arXiv preprint"},{"key":"ref73","first-page":"1","article-title":"Griffon v2: Advancing multimodal perception with high-resolution scaling and visual-language co-referring","author":"Zhan","year":"2024","journal-title":"arXiv preprint"},{"key":"ref74","article-title":"Ferretv 2: An improved baseline for referring and grounding with large language models","author":"Zhang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref75","article-title":"Mllms know where to look: Training-free perception of small visual details with multimodal 11 ms","volume-title":"Proceedings of International Conference on Learning Representations (ICLR)","author":"Zhang"},{"key":"ref76","article-title":"Internlmxcomposer: A vision-language large model for advanced text-image comprehension and composition","author":"Zhang","year":"2023","journal-title":"arXiv preprint"},{"key":"ref77","article-title":"Opt: Open pre-trained transformer language models","author":"Zhang","year":"2022","journal-title":"arXiv preprint"},{"key":"ref78","first-page":"2","article-title":"Llavar: Enhanced visual instruction tuning for text-rich image understanding","author":"Zhang","year":"2023","journal-title":"arXiv preprint"},{"key":"ref79","first-page":"5","article-title":"Mme-realworld: Could your multimodal 11 m challenge high-resolution real-world scenarios that are difficult for humans?","author":"Zhang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref80","article-title":"Minigpt-4: Enhancing vision-language understanding with advanced large language models","author":"Zhu","year":"2023","journal-title":"arXiv preprint"}],"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\/11445959.pdf?arnumber=11445959","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T05:03:38Z","timestamp":1777611818000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11445959\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,19]]},"references-count":80,"URL":"https:\/\/doi.org\/10.1109\/iccv51701.2025.02238","relation":{},"subject":[],"published":{"date-parts":[[2025,10,19]]}}}