{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,17]],"date-time":"2026-07-17T16:13:34Z","timestamp":1784304814977,"version":"3.55.0"},"reference-count":58,"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":["92470121,62402016"],"award-info":[{"award-number":["92470121,62402016"]}],"id":[{"id":"10.13039\/501100001809","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\/iccv51701.2025.01620","type":"proceedings-article","created":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T19:45:49Z","timestamp":1777491949000},"page":"17443-17453","source":"Crossref","is-referenced-by-count":7,"title":["OCR Hinders RAG: Evaluating the Cascading Impact of OCR on Retrieval-Augmented Generation"],"prefix":"10.1109","author":[{"given":"Junyuan","family":"Zhang","sequence":"first","affiliation":[{"name":"Shanghai AI Laboratory"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Qintong","family":"Zhang","sequence":"additional","affiliation":[{"name":"Shanghai AI Laboratory"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bin","family":"Wang","sequence":"additional","affiliation":[{"name":"Shanghai AI Laboratory"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Linke","family":"Ouyang","sequence":"additional","affiliation":[{"name":"Shanghai AI Laboratory"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zichen","family":"Wen","sequence":"additional","affiliation":[{"name":"Shanghai AI Laboratory"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ying","family":"Li","sequence":"additional","affiliation":[{"name":"Beihang University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ka-Ho","family":"Chow","sequence":"additional","affiliation":[{"name":"The University of HongKong"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Conghui","family":"He","sequence":"additional","affiliation":[{"name":"Shanghai AI Laboratory"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wentao","family":"Zhang","sequence":"additional","affiliation":[{"name":"Peking University"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Gpt-4 technical report","author":"Achiam","year":"2023","journal-title":"arXiv preprint"},{"key":"ref2","article-title":"Mc-llava: Multi-concept personalized vision-language model","author":"An","year":"2024","journal-title":"arXiv preprint"},{"key":"ref3","article-title":"Unictokens: Boosting personalized understanding and generation via unified concept tokens","author":"An","year":"2025","journal-title":"arXiv preprint"},{"key":"ref4","article-title":"Qwen2. 5-vl technical report","author":"Bai","year":"2025","journal-title":"arXiv preprint"},{"key":"ref5","article-title":"Nougat: Neural optical understanding for academic documents","volume-title":"The Twelfth International Conference on Learning Representations","author":"Blecher","year":"2024"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v38i16.29728"},{"key":"ref7","article-title":"Bge m3-embedding: Multi-lingual, multifunctionality, multi-granularity text embeddings through selfknowledge distillation","author":"Chen","year":"2024","journal-title":"arXiv preprint"},{"key":"ref8","article-title":"Expanding performance boundaries of open-source multimodal models with model, data, and testtime scaling","author":"Chen","year":"2024","journal-title":"arXiv preprint"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1007\/s11432-024-4231-5"},{"key":"ref10","article-title":"M3docrag: Multi-modal retrieval is what you need for multi-page multi-document understanding","author":"Cho","year":"2024","journal-title":"arXiv preprint"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.findings-emnlp.161"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.acl-long.57"},{"key":"ref13","article-title":"The llama 3 herd of models","author":"Dubey","year":"2024","journal-title":"arXiv preprint"},{"key":"ref14","article-title":"Ragas: Automated evaluation of retrieval augmented generation","author":"Es","year":"2023","journal-title":"arXiv preprint"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.acl-long.540"},{"key":"ref16","article-title":"Colpali: Efficient document retrieval with vision language models","author":"Faysse","year":"2024","journal-title":"arXiv preprint"},{"key":"ref17","article-title":"Retrieval-augmented generation for large language models: A survey","author":"Gao","year":"2023","journal-title":"arXiv preprint"},{"key":"ref18","article-title":"Cuad: An expert-annotated nlp dataset for legal contract review","author":"Hendrycks","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref19","article-title":"spaCy 2: Natural language understanding with Bloom embeddings, convolutional neural networks and incremental parsing","author":"Honnibal","year":"2017","journal-title":"To appear"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.52202\/079017-2145"},{"key":"ref21","article-title":"Gpt-4o system card","author":"Hurst","year":"2024","journal-title":"arXiv preprint"},{"key":"ref22","article-title":"Financebench: A new benchmark for financial question answering","author":"Islam","year":"2023","journal-title":"arXiv preprint"},{"issue":"251","key":"ref23","first-page":"1","article-title":"Atlas: Few-shot learning with retrieval augmented language models","volume":"24","author":"Izacard","year":"2023","journal-title":"Journal of Machine Learning Research"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-86337-1_27"},{"key":"ref25","first-page":"94599474","article-title":"Retrievalaugmented generation for knowledge-intensive nlp tasks","volume":"33","author":"Lewis","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref26","article-title":"Towards general text embeddings with multi-stage contrastive learning","author":"Li","year":"2023","journal-title":"arXiv preprint"},{"key":"ref27","article-title":"Readoc: A unified benchmark for realistic document structured extraction","author":"Li","year":"2024","journal-title":"arXiv preprint"},{"key":"ref28","article-title":"Deepseek-v3 technical report","author":"Liu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref29","article-title":"Focus anywhere for finegrained multi-page document understanding","author":"Liu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref30","article-title":"Shifting ai efficiency from model-centric to data-centric compression","author":"Liu","year":"2025","journal-title":"arXiv preprint"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1007\/s11432-024-4235-6"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-73414-4_14"},{"key":"ref33","article-title":"Mmlongbench-doc: Benchmarking long-context document understanding with visualizations","author":"Ma","year":"2024","journal-title":"arXiv preprint"},{"key":"ref34","article-title":"Nvretriever: Improving text embedding models with effective hard-negative mining","author":"de Souza P Moreira","year":"2024","journal-title":"arXiv preprint"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52734.2025.02313"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1007\/978-1-4020-6754-9_9881"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1007\/3-540-58338-6_63"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1162\/tacl_a_00605"},{"key":"ref39","first-page":"109","article-title":"Okapi at trec-3","volume":"109","author":"Stephen","year":"1995","journal-title":"Nist Special Publication Sp"},{"key":"ref40","article-title":"Ares: An automated evaluation framework for retrieval-augmented generation systems","author":"Saad-Falcon","year":"2023","journal-title":"arXiv preprint"},{"key":"ref41","article-title":"Ext2gen: Alignment through unified extraction and generation for robust retrieval-augmented generation","author":"Song","year":"2025","journal-title":"arXiv preprint"},{"key":"ref42","article-title":"Multihop-rag: Benchmarking retrieval-augmented generation for multi-hop queries","author":"Tang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref43","article-title":"Novelseek: When agent becomes the scientist-building closed-loop system from hypothesis to verification","volume-title":"arXiv preprint","author":"Team","year":"2025"},{"key":"ref44","article-title":"Qwen Team","year":"2024","journal-title":"Qwen2.5: A party of foundation models"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1145\/2682862.2682863"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.01789"},{"key":"ref47","article-title":"Unimernet: A universal network for real-world mathematical expression recognition","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref48","article-title":"Mineru: An open-source solution for precise document content extraction","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref49","article-title":"Qwen2-vl: Enhancing vision-language model\u2019s perception of the world at any resolution","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref50","article-title":"General ocr theory: Towards ocr-2.0 via a unified end-to-end model","author":"Wei","year":"2024","journal-title":"arXiv preprint"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.findings-acl.802"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.emnlp-main.505"},{"key":"ref53","article-title":"How faithful are rag models? quantifying the tug-of-war between rag and 11 ms \u2019 internal prior","author":"Wu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.acl-long.9"},{"key":"ref55","article-title":"Qwen2 technical report","author":"Yang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref56","article-title":"Visrag: Vision-based retrieval-augmented generation on multi-modality documents","author":"Yu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref57","article-title":"Document parsing unveiled: Techniques, challenges, and prospects for structured information extraction","author":"Zhang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref58","article-title":"Docbench: A benchmark for evaluating 11 m -based document reading systems","author":"Zou","year":"2024","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\/11445619.pdf?arnumber=11445619","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T05:02:22Z","timestamp":1777611742000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11445619\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,19]]},"references-count":58,"URL":"https:\/\/doi.org\/10.1109\/iccv51701.2025.01620","relation":{},"subject":[],"published":{"date-parts":[[2025,10,19]]}}}