Papers for

digital archive managers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Domain adaptation improves handwritten character and text recognition accuracy

Domain Adaptation Based Pipeline for Character Classification and Handwritten Text Recognition

Abstract: Character classification and handwritten text recognition, particularly in specialized tasks, often suffer from limited labeled data, while a large amount of unlabeled data remains unused. To address this challenge, we propose a novel unsupervised domain adaptation-based pipeline to reduce the distribution gap between source and target domains. Our pipeline follows three main steps. In the first step, we pretrained an expert model on source data, to learn a strong general representation. In the second step, we perform unsupervised domain adaptation to leverage unlabeled target data, adapting the model to the specific characteristics of the target domain. In the third step, we trained the expert model on source data in the target style. By aligning the feature distributions between the source and target, our method enables the model to generalize effectively across diverse datasets. The unsupervised training increases the computation time only during training not in inference. We evaluate the pipeline on both character-level and historical handwritten line recognition tasks, demonstrating its flexibility. We establish a new state-of-the-art on the Safran-MNIST-DLS dataset defined in Track 2 of the DAGECC competition with 76.17% Macro F1-score and achieve competitive results on each of the specific datasets of READ 2018 without using the labels of the target data. Code is available at https://github.com/simon-corbi/DRANet-for-classification-and-htr

Fri 13 NovComputer Vision and Pattern RecognitionMachine Learning
The gist
Handwritten character recognition can struggle when there isn’t enough labeled training data for specific handwriting styles. The authors developed a method that helps computers learn from lots of unlabeled handwriting samples by adjusting a model trained on labeled data to better fit the new style. This method helps the model perform well across different handwriting styles without needing labeled examples for each one. Their approach improves accuracy on challenges involving historical and specialized handwriting datasets.
Open → 2611.17952v1

Redundant facts in long documents challenge standard citation scoring

One Fact, Many Places: Redundant Evidence in Long Visual Documents Breaks Single-Gold Evaluation of Retrieval, Citation and Attribution

Abstract: Document understanding is increasingly judged on where an answer comes from, scored against annotated gold locations as if each fact lived in one place. In long visual documents, a fact often lives in several places: a chart value is restated in the text, a key figure reappears in the summary. We ask how often such copies occur in standard benchmarks and what they do to evaluation. We annotate 420 questions from MMLongBench-Doc and MMDocRAG with the locations that independently establish each required fact, checked by two independent reviewers. Copies are common and often cross-modal: 19.5% of MMLongBench-Doc test questions have a strict copy outside the gold, and in 28% of MMDocRAG questions a non-gold candidate alone supplies a required fact. Single-gold scoring then fails in three predictable ways. A retriever that finds a copy is scored as a miss: page recall@1 of text and visual retrievers is underestimated by 13-24 points on questions with a cross-page copy. A model that cites a copy is scored as citing wrongly: crediting equivalent evidence reverses 11% of pairwise orderings among 65 competitive MMDocRAG systems. Masking the gold region leaves the fact available elsewhere: with a copy in the input, 41-56% of correct answers across four models survive it yet fail once every annotated instance is masked. Set-level masking with automatically found copies closes 78% of the gap to masking every annotated instance at matched granularity (three models); an unverified key-string rule lowers survival at least as far but masks three times the area. We release the annotations and redundancy-aware metrics.

Mon 19 OctComputer Vision and Pattern Recognition
The gist
When computers try to find where a fact comes from in long documents with text and pictures, they usually expect that each fact appears in only one place. But in reality, the same fact can appear multiple times, such as in charts and text. The authors showed that this redundancy happens often and causes current systems to be unfairly penalized for finding correct facts in different places. They created new annotations and scoring methods that account for these repeated facts, improving how we evaluate fact retrieval and citation.
Open → 2610.26401v1

Dense retrievers struggle with mixed text and image document searches

Chaos in the Text: Revealing the Modality Preference in Mixed-Modality Retrievers

Abstract: Dense retrievers have made significant progress on text and image corpora, but whether these capabilities extend reliably to mixed corpora containing text, image, and fused text-image documents remains unclear. In this paper, we systematically examine retrievers across architectures and find that their performance is highly sensitive to modality composition. As image documents are progressively replaced with semantically corresponding text representations, retrieval performance follows a pronounced V-shaped curve, remaining strong on single-modality corpora but degrading substantially when modalities coexist. In particular, irrelevant text causes more severe degradation than an equal number of irrelevant images, a phenomenon we term Chaos in the Text. Further analysis reveals modality preference, whereby text representations receive systematically higher similarity scores, allowing irrelevant text to outrank relevant images. To mitigate this bias, we introduce Trident, which constructs text, image, and fused text-image views of each document as co-equal positives and jointly optimizes relevance discrimination and positive-view balance through Multi-Positive View InfoNCE. Experiments across visual document and natural image benchmarks show that trident improves mixed-modality retrieval on both CLIP-based and VLM-based architectures, reduces sensitivity to modality composition and text distractors, and increases average single-modality retrieval performance.

Thu 8 OctInformation Retrieval
The gist
Finding relevant information in mixed collections of text and images is tricky for current computer systems. The authors show that these systems prefer text and can get confused when both text and images are present. This confusion leads to poorer search results, especially when irrelevant text distracts the system more than irrelevant images. They propose a new method called Trident to balance the importance of text and images, improving search accuracy in mixed collections.
Open → 2610.11816v1

Small model predicts which images aid faster and accurate answers

Decide Before You Look: Learning Which Retrieved Memories Deserve Pixels

Abstract: Multimodal assistants answer questions from long-term memories that contain images. After retrieval, each retrieved image reaches the answering model either as pixels, at about a thousand visual tokens per image, or as a stored text proxy that often misses the detail the question asks about. We find that the benefit of pixels usually comes from one or two retrieved memories, and that it can be predicted before the answering model runs, without reading any full-resolution image. In PixelTriage, a plug-in placed after retrieval, a small model that does not generate text reads the dialogue, a short note and a thumbnail of each retrieved memory and predicts how much its pixels would add. It is trained on synthetic memory episodes labeled by a frozen 27B model that answers each question with and without each memory's pixels. With a 7B answering model, PixelTriage lies on the accuracy--cost frontier of M$^3$Exam, DMV and MemEye and uses 11--23\% of the visual tokens without a significant loss of accuracy. On DMV it answers 2.9 times faster than opening all images. It outperforms retrieval order and uniform down-sizing at equal budgets and transfers to other memory systems and to a 397B answering model.

Tue 6 OctComputer Vision and Pattern RecognitionArtificial Intelligence
The gist
Answering questions with many images can be slow because looking at each full image uses a lot of computer power. The authors show that usually only one or two images really help with the answer. They created PixelTriage, a small model that looks at tiny picture previews and some text to decide which images are worth loading in full detail. This method can speed up answering by almost three times without losing accuracy and works well across different memory systems and models.
Open → 2610.07984v1

Multi-vector visual document search sped up with smaller query models

ColNanoVDR: Document-Free Query Distillation for Multi-Vector Visual Document Retrieval via Optimal Transport

Abstract: Multi-vector retrievers built on vision-language models lead visual document retrieval (VDR), but they run a multi-billion-parameter query encoder on every search. Distilling this encoder into a small student that queries the teacher's existing index would remove the bottleneck. The standard recipe, however, matches the teacher's MaxSim scores and so requires encoding and caching every training page, which can reach terabytes of page tokens. NanoVDR avoids pages entirely by training on the teacher's query embeddings alone, but only for single-vector retrievers. We present ColNanoVDR, to our knowledge the first framework to bring this document-free distillation to multi-vector VDR. Its objective, OTW (Optimal Transport with Learned Weights), aligns the student's query tokens with the teacher's by entropic optimal transport, with a learned weight for each student token, and needs no correspondence between the two tokenizations. We prove that the resulting alignment cost bounds the MaxSim score difference on every page. Distilled from five state-of-the-art teachers, the 149M text-only students retain about 95% of their teachers' NDCG@5 on ViDoRe v1-v3 while encoding queries up to 26x faster. Under identical training, OTW matches score distillation while encoding no page and reading 12.6x less cached teacher data.

Mon 28 SeptInformation RetrievalComputation and LanguageComputer Vision and Pattern Recognition
The gist
Searching large collections of visual documents is slow because the computer runs a huge model every time. The authors developed a way to teach a smaller, faster model to ask the big model for help without needing to look at all the documents. They use a math technique called optimal transport to match parts of the smaller model’s questions to the big model’s answers. This method nearly keeps the search accuracy while making query processing up to 26 times faster.
Open → 2609.34899v1

Historical persian manuscript dataset improves word spotting accuracy

HPMD: A Historical Persian Manuscript Dataset for Word Spotting with Line-Level Annotation

Abstract: Large collections of historical Persian manuscripts have been digitized, but searching them is still slow and mostly manual. Historians usually want to find where a specific name, date, event, or topic appears, which is a word spotting problem. Progress on this task is limited by two things. First, there is almost no public dataset of historical Persian handwriting; the only notable resource, OpenITI MAKHZAN, contains a relatively small Persian portion. Second, word spotting models usually need word-level bounding boxes, which are very expensive to annotate. In this paper we introduce a new dataset of 223 pages, 3,678 lines, 37,631 words, and 130,630 characters, collected from diverse historical Persian books of poetry and prose and annotated at the region, line, and text level. We also propose a baseline that is trained only with line-level annotations but returns word-level locations. A fine-tuned line detector finds text lines, and a fine-tuned CRNN recognizer trained with CTC produces a frame-by-character posterior matrix for each line. Instead of decoding the most probable character at each frame, the query is scored directly against this matrix, so visually similar characters in Persian such as be and pe no longer cause hard failures. The frame alignment also gives the horizontal position of the word inside the line. On the test set, the fine-tuned line detector reaches an F1 of 0.892, and posterior-based search raises the word spotting F1 from 0.487 to 0.558 compared with exact matching on the decoded text, with the decision threshold selected on a held-out validation set. A PHOC attribute-embedding baseline that additionally receives oracle word boundaries at test time reaches an F1 of 0.449, below the proposed method. We also report a distributional analysis of the dataset, a taxonomy of retrieval errors, and a per-conditionbreakdown of performance.

Mon 28 SeptComputer Vision and Pattern Recognition
The gist
Searching old Persian handwritten books for specific words is hard because there are no good public datasets and marking words manually is expensive. The authors created a new large dataset of Persian manuscripts with detailed line and text labels. They also developed a method that finds lines and spots words without needing word-level marking during training, improving search accuracy. This approach handles visually similar Persian letters better, making it easier to locate words in the text.
Open → 2609.34490v1

Visual document retrieval improves by adapting queries with residual feedback

Test-Time Adaptation with Query-Dependent Residuals for Visual Document Retrieval

Abstract: Visual document retrieval (VDR) systems depend on page embeddings computed before deployment, which makes adaptation difficult when encoder parameters or corpus re-encoding are unavailable. Rerankers provide useful relevance signals, but conventional reranking applies them only to selected queries and candidate pages. We introduce Q-REACT, a query-side test-time adaptation method that converts limited reranker feedback into reusable retrieval improvements. Q-REACT learns a shared low-rank transformation that produces query-dependent residuals, combines adapted query scores with document-level context, and distills reranker preferences with a student distribution normalized over the complete task-specific page index. This design lets unscored pages compete through cached embeddings while keeping the encoders and page index fixed. Across eight ViDoRe V3 tasks and five open-weight and proprietary backbones, Q-REACT improves average retrieval over evaluated baselines at sparse and full-coverage budgets, transfers to held-out queries and tasks, and adds little inference overhead. The results show that finite reranker feedback can be amortized across a query collection without retraining or rebuilding the retriever.

Wed 23 SeptInformation Retrieval
The gist
Visual document retrieval systems find relevant pages in large document collections using precomputed page data, but adapting them after deployment is hard. The authors present Q-REACT, a method that uses feedback from a secondary ranking process to adjust queries without changing the underlying document data. This approach allows all documents to compete fairly during retrieval and works well across multiple tasks and retrieval systems. It improves performance with minimal extra computing time and does not require changing or re-encoding the documents.
Open → 2609.27688v1

Query adaptive indexing improves retrieval in expert archives

ORDER: Task-Conditioned Routing for Retrieval-Augmented Generation

Abstract: Retrieval-Augmented Generation (RAG) pipelines typically rely on a fixed indexing and retrieval configuration determined at preprocessing time. This one-size-fits-all design is ill-suited to domain-expert settings, where heterogeneous queries require different chunking granularities, metadata constraints, and source-selection strategies. As a result, configurations that are effective for one family of queries often perform poorly for others. In this paper, we introduce ORDER (Optimal Routing for Dynamic Evidence Retrieval), a query-conditioned RAG framework that jointly adapts indexing and retrieval to the incoming query. Our approach first discovers semantic clusters over a given set of questions associated to a corpus and learns, for each cluster, a chunking strategy together with a suited metadata filtering and reranking configuration. At inference time, queries are routed to the appropriate pre-built index through nearest-centroid assignment. To further improve retrieval, we propose a supervised query router (QRe) that predicts which collections are most likely to contain relevant evidence, coupled with a Uniform Multi-source Sampler (UMS) that allocates the retrieval budget evenly across the selected sources. We evaluate our framework on large-scale, heterogeneous historical archives and show that conditioning both indexing and retrieval on the query consistently outperforms both naive baselines and strong state-of-the-art RAG systems in complex expert-domain environments.

Tue 15 SeptArtificial Intelligence
The gist
Many systems that find and generate answers from large collections use the same setup for all questions, even though different questions need different ways of organizing and searching the data. The authors propose ORDER, a method that changes how the data is divided and searched depending on the type of question asked. First, it groups similar questions and learns the best way to split and filter the underlying information for each group. Then, when a new question comes in, it routes it to the right way of searching based on its similarity to the groups. This approach helps get better answers from complex, historical archives with diverse information.
Open → 2609.17012v1