{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T10:24:23Z","timestamp":1777890263989,"version":"3.51.4"},"reference-count":91,"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\/iccv51701.2025.01648","type":"proceedings-article","created":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T19:45:49Z","timestamp":1777491949000},"page":"17739-17750","source":"Crossref","is-referenced-by-count":0,"title":["Harmonizing Visual Representations for Unified Multimodal Understanding and Generation"],"prefix":"10.1109","author":[{"given":"Size","family":"Wu","sequence":"first","affiliation":[{"name":"Nanyang Technological University,S-Lab"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wenwei","family":"Zhang","sequence":"additional","affiliation":[{"name":"Shanghai AI Laboratory Research"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Lumin","family":"Xu","sequence":"additional","affiliation":[{"name":"The Chinese University of Hong Kong"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Sheng","family":"Jin","sequence":"additional","affiliation":[{"name":"SenseTime Research and Tetras. AI"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zhonghua","family":"Wu","sequence":"additional","affiliation":[{"name":"SenseTime Research"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Qingyi","family":"Tao","sequence":"additional","affiliation":[{"name":"SenseTime Research"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wentao","family":"Liu","sequence":"additional","affiliation":[{"name":"SenseTime Research and Tetras. AI"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wei","family":"Li","sequence":"additional","affiliation":[{"name":"Nanyang Technological University,S-Lab"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Chen Change","family":"Loy","sequence":"additional","affiliation":[{"name":"Nanyang Technological University,S-Lab"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Meissonic: Revitalizing masked generative transformers for efficient high-resolution text-to-image synthesis","volume-title":"The Thirteenth International Conference on Learning Representations","author":"Bai","year":"2024"},{"key":"ref2","article-title":"Beit: Bert pre-training of image transformers","author":"Bao","year":"2021","journal-title":"arXiv preprint arXiv"},{"issue":"3","key":"ref3","article-title":"Improving image generation with better captions","volume-title":"Computer Science","volume":"2","author":"Betker","year":"2023"},{"key":"ref4","article-title":"Language models are few-shot learners","author":"Brown","year":"2020","journal-title":"arXiv preprint arXiv"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01103"},{"key":"ref6","article-title":"Muse: Text-to-image generation via masked generative transformers","author":"Chang","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1109\/WACV.2018.00097"},{"key":"ref8","article-title":"PixArt-alpha: Fast training of diffusion transformer for photorealistic text-to-image synthesis","author":"Chen","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref9","article-title":"Pixart-alpha: Fast training of diffusion transformer for photorealistic text-to-image synthesis","author":"Chen","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-73411-3_5"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-72643-9_22"},{"key":"ref12","article-title":"Microsoft coco captions: Data collection and evaluation server","author":"Chen","year":"2015","journal-title":"arXiv preprint arXiv"},{"key":"ref13","article-title":"Janus-pro: Unified multimodal understanding and generation with data and model scaling","author":"Chen","year":"2025","journal-title":"arXiv preprint arXiv"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1007\/s11432-024-4231-5"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52733.2024.02283"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.00276"},{"key":"ref17","article-title":"Mobilevlm: A fast, reproducible and strong vision language assistant for mobile devices","author":"Chu","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref18","article-title":"Mobilevlm v2: Faster and stronger baseline for vision language model","author":"Chu","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref19","volume-title":"Instructblip: Towards general-purpose vision-language models with instruction tuning","author":"Dai","year":"2023"},{"key":"ref20","volume-title":"Laion-aesthetics-umap","year":"2022"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR46437.2021.01268"},{"key":"ref23","author":"Esser","year":"2024","journal-title":"Scaling rectified flow transformers for high-resolution image synthesis"},{"key":"ref24","article-title":"Fluid: Scaling autoregressive text-to-image generative models with continuous tokens","author":"Fan","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref25","article-title":"Mme: A comprehensive evaluation benchmark for multimodal large language models","author":"Fu","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref26","article-title":"Seed-x: Multimodal models with unified multi-granularity comprehension and generation","author":"Ge","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.52202\/075280-2270"},{"key":"ref28","article-title":"Infinity-mm: Scaling multimodal performance with large-scale and high-quality instruction data","author":"Gu","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01553"},{"key":"ref30","article-title":"Classifier-free diffusion guidance","author":"Ho","year":"2022","journal-title":"arXiv preprint arXiv"},{"key":"ref31","first-page":"6840","article-title":"Denoising diffusion probabilistic models","volume":"33","author":"Ho","year":"2020","journal-title":"Advances in neural information processing systems"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00686"},{"key":"ref33","volume-title":"text-to-image-2m","year":"2024"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.00976"},{"key":"ref35","volume-title":"Auto-encoding variational bayes","author":"Kingma","year":"2013"},{"key":"ref36","article-title":"Seed-bench: Benchmarking multimodal llms with generative comprehension","author":"Li","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref37","article-title":"Llava-onevision: Easy visual task transfer","author":"Li","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref38","article-title":"Playground v2. 5: Three insights towards enhancing aesthetic quality in text-to-image generation","author":"Li","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref39","article-title":"Autoregressive image generation without vector quantization","author":"Li","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref40","article-title":"Densefusion-1m: Merging vision experts for comprehensive multimodal perception","author":"Li","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.emnlp-main.20"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1109\/tpami.2025.3637265"},{"key":"ref43","article-title":"Dual diffusion for unified image generation and understanding","author":"Li","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref44","article-title":"Hunyuan-DiT: A powerful multi-resolution diffusion transformer with fine-grained chinese understanding","author":"Li","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02484"},{"key":"ref46","article-title":"Visual instruction tuning","volume":"36","author":"Liu","year":"2024","journal-title":"Advances in neural information processing systems"},{"key":"ref47","article-title":"World model on million-length video and language with ringattention","author":"Liu","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-72658-3_13"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01167"},{"key":"ref50","article-title":"Deepseek-vl: towards real-world vision-language understanding","author":"Lu","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref51","article-title":"Open-magvit2: An open-source project toward democratizing auto-regressive visual generation","author":"Luo","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref52","volume-title":"Megalith-huggingface","year":"2024"},{"key":"ref53","article-title":"Public domain 12m: A highly aesthetic image-text dataset with novel governance mechanisms","author":"Meyer","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref54","article-title":"Journeydb: A benchmark for generative image understanding","author":"Pan","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.00387"},{"key":"ref56","article-title":"SDXL: Improving latent diffusion models for high-resolution image synthesis","author":"Podell","year":"2024","journal-title":"ICLR"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52734.2025.00243"},{"key":"ref58","volume-title":"Improving language understanding by generative pre-training","author":"Radford","year":"2018"},{"issue":"8","key":"ref59","article-title":"Language models are unsupervised multitask learners","volume-title":"OpenAI blog","volume":"1","author":"Radford","year":"2019"},{"key":"ref60","first-page":"8748","article-title":"Learning transferable visual models from natural language supervision","volume-title":"International conference on machine learning","author":"Radford","year":"2021"},{"key":"ref61","first-page":"8821","article-title":"Zero-shot text-to-image generation","volume-title":"International conference on machine learning","author":"Ramesh","year":"2021"},{"issue":"2","key":"ref62","article-title":"Hierarchical text-conditional image generation with clip latents","volume":"1","author":"Ramesh","year":"2022","journal-title":"arXiv preprint arXiv"},{"key":"ref63","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01042"},{"key":"ref64","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01042"},{"key":"ref65","article-title":"Llamafusion: Adapting pretrained language models for multimodal generation","author":"Shi","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref66","article-title":"From pixels to prose: A large dataset of dense image captions","author":"Singla","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref67","article-title":"Autoregressive model beats diffusion: Llama for scalable image generation","author":"Sun","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref68","article-title":"Generative pretraining in multimodality","author":"Sun","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref69","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.01365"},{"key":"ref70","article-title":"Chameleon: Mixed-modal early-fusion foundation models","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref71","article-title":"Gemini: a family of highly capable multimodal models","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref72","article-title":"Metamorph: Multimodal understanding and generation via instruction tuning","author":"Tong","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref73","article-title":"Llama 2: Open foundation and fine-tuned chat models","author":"Touvron","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref74","article-title":"Neural discrete representation learning","volume":"30","author":"Van Den Oord","year":"2017","journal-title":"Advances in neural information processing systems"},{"key":"ref75","article-title":"Illume: Illuminating your llms to see, draw, and self-enhance","author":"Wang","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref76","article-title":"Qwen2-vl: Enhancing vision-language model\u2019s perception of the world at any resolution","author":"Wang","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref77","article-title":"Emu3: Next-token prediction is all you need","author":"Wang","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref78","article-title":"Janus: Decoupling visual encoding for unified multimodal understanding and generation","author":"Wu","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref79","article-title":"F-Imm: Grounding frozen large multimodal models","author":"Wu","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref80","article-title":"Vila-u: a unified foundation model integrating visual understanding and generation","author":"Wu","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref81","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52734.2025.01241"},{"key":"ref82","article-title":"Show-o: One single transformer to unify multimodal understanding and generation","author":"Xie","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref83","article-title":"Qwen2. 5 technical report","author":"Yang","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref84","article-title":"Language model beats diffusion-tokenizer is key to visual generation","author":"Yu","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref85","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.00913"},{"key":"ref86","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-024-02214-4"},{"key":"ref87","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.01100"},{"key":"ref88","article-title":"Monoformer: One transformer for both diffusion and autoregression","author":"Zhao","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref89","article-title":"Transfusion: Predict the next token and diffuse images with one multi-modal model","author":"Zhou","year":"2024","journal-title":"arXiv preprint arXiv"},{"key":"ref90","article-title":"Minigpt-4: Enhancing vision-language understanding with advanced large language models","author":"Zhu","year":"2023","journal-title":"arXiv preprint arXiv"},{"key":"ref91","article-title":"Lumina-Next: Making Lumina-T2X stronger and faster with Next-DiT","author":"Zhuo","year":"2024","journal-title":"arXiv preprint arXiv"}],"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\/11446141.pdf?arnumber=11446141","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T05:33:23Z","timestamp":1777613603000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11446141\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,19]]},"references-count":91,"URL":"https:\/\/doi.org\/10.1109\/iccv51701.2025.01648","relation":{},"subject":[],"published":{"date-parts":[[2025,10,19]]}}}