Skip to content
AI News HubLIVE
Original source3 min read

Evaluation protocols for historical script classification: the cost of manuscript-level leakage and the role of soft labels

Summary

A new arXiv paper quantifies how page-level leakage inflates accuracy in historical script classification (97% to 54%) and explores expert soft labels for better calibration and alignment with expert assessment, offering evaluation protocol recommendations.

SourcearXiv Computer VisionAuthor: Daniil I. Naryzhnyi (ITMO University), Roman A. Popkov (ITMO University), Daniil A. Bakalin (ITMO University)
Evaluation protocols for historical script classification: the cost of manuscript-level leakage and the role of soft labels
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 7 Nov 2026]

Title:Evaluation protocols for historical script classification: the cost of manuscript-level leakage and the role of soft labels

View a PDF of the paper titled Evaluation protocols for historical script classification: the cost of manuscript-level leakage and the role of soft labels, by Daniil I. Naryzhnyi (ITMO University) and 2 other authors

View PDF

Abstract:A paleographer applies a script type classifier to unseen manuscripts. We measure how much testing on other pages of training manuscripts inflates accuracy. On VML-HP-ext (715 pages, 171 Hebrew manuscripts, 14 script classes) we compare page-level and manuscript-level splits. Page-level accuracy drops from 97% to 54% (59% at manuscript level), and is 39% on the official blind test set. The gap of 40 to 47 percentage points (pp) at patch, page and manuscript level is statistically significant and reproduces with ResNet18; on binarized patches (ink and background only) it is 34 to 42 pp, also significant. On unseen manuscripts the model is overconfident: mean confidence 70% against 54% correct pages; patch-level expected calibration error (ECE, mean gap between confidence and accuracy) rises from 0.065 to 0.444. Binarization leaves patch-level overconfidence but cuts it markedly at page level. Confidence still helps triage: in the more confident half of pages 67% are correct, 41% in the rest. Training on expert soft labels showed no evidence of an accuracy difference, though a small page-level loss cannot be ruled out. The predicted distributions are closer to the expert's assessment. ECE is lower at patch level (0.21 to 0.31) and in most runs at page level, but not significantly at manuscript level. Label smoothing (blurring the label, no expert input) also lowers ECE but comes no closer to the expert. Temperature scaling (post-training confidence correction) helps significantly only at patch level. On 97 pages of 16 manuscripts with mixed expert assessment, soft labels cut divergence from the expert by 22 to 30% (significant in four of five runs); the point estimate of accuracy is 16.5 to 23.7 pp below hard labels, but with 16 manuscripts the loss can neither be confirmed nor ruled out. We propose recommendations for the evaluation protocol; code will be released with the paper

Comments: 29 pages, 8 tables, 2 figures

Subjects:

Computer Vision and Pattern Recognition (cs.CV)

Cite as: arXiv:2611.12969 [cs.CV]

(or arXiv:2611.12969v1 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2611.12969

arXiv-issued DOI via DataCite

Submission history

From: Daniil Naryzhnyi [view email] [v1] Sat, 7 Nov 2026 16:37:46 UTC (567 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Evaluation protocols for historical script classification: the cost of manuscript-level leakage and the role of soft labels, by Daniil I. Naryzhnyi (ITMO University) and 2 other authors

View PDF

view license

Additional Features

Audio Summary

Current browse context:

cs.CV

new | recent | 2026-11

Change to browse by:

cs

References & Citations

SciX

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

InvestorsAdvanced

Key points

  • Page-level accuracy drops from 97% to 54% (59% at manuscript level) and 39% on blind test, a significant 40-47 pp gap.
  • Models are overconfident on unseen manuscripts; patch-level ECE rises from 0.065 to 0.444; binarization reduces page-level overconfidence.
  • Expert soft labels show no clear accuracy gain but bring predictions closer to expert assessment and lower ECE and divergence.
  • Temperature scaling helps calibration significantly only at patch level; authors propose protocol recommendations and will release code.

Highlights and analysis are generated automatically and may contain errors. Check the original source.