Line and Word Level Annotations for Devanagari Sanskrit Manuscripts (Shirorekha Segmentation)

Read full text Summarise with VIRAT
Authors: Latik Kothwar, Jagruti S. Dehankar
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Shirorekha Annotations for Devanagari Manuscripts (125 Folios) Dataset Overview This dataset provides pixel-wise line masks (Shirorekha annotations) for 125 selected scanned Devanagari manuscript folios. These annotations are specifically designed to support line-level segmentation and historical manuscript analysis. The original images are sourced from the PE-OCR Sanskrit benchmark dataset introduced by:Maheshwari et al. (2022). "Findings of EMNLP." Original Source: https://github.com/ayushbits/pe-ocr-sanskrit Annotation Details - Target: Shirorekha (header line) of Devanagari characters.- Format: Binary PNG masks.- Pixel Values: 255 (Foreground/Shirorekha), 0 (Background).- Process: Each folio was manually annotated by two Sanskrit paleography experts, with a third expert resolving any disagreements. Data Split The 125 folios are partitioned as follows:- Training: 85 folios- Validation: 15 folios- Test: 25 foliosNote: While data augmentation is recommended for training, these provided masks are the original, non-augmented annotations. Filename Mapping Table To maintain consistency, the original images from the PE-OCR source were renamed. The table below provides the mapping between the original source directory/number and the renamed filename in this annotated dataset. Sr. | Source DS | Page Range | Orig. No | Renamed Name----|-----------|-----------------|----------|-------------1 | BHSv1S1 | Page-001-102 | 6 | img-0012 | BHSv1S1 | Page-001-102 | 8 | img-0023 | BHSv1S1 | Page-001-102 | 9 | img-0034 | BHSv1S1 | Page-001-102 | 10 | img-0045 | BHSv1S1 | Page-001-102 | 21 | img-0056 | BHSv1S1 | Page-001-102 | 29 | img-0067 | BHSv1S1 | Page-001-102 | 32 | img-0078 | BHSv1S1 | Page-001-102 | 37 | img-0089 | BHSv1S1 | Page-001-102 | 42 | img-00910 | BHSv1S1 | Page-001-102 | 43 | img-01011 | BHSv1S1 | Page-001-102 | 44 | img-01112 | BHSv1S1 | Page-001-102 | 47 | img-01213 | BHSv1S1 | Page-001-102 | 49 | img-01314 | BHSv1S1 | Page-001-102 | 51 | img-01415 | BHSv1S1 | Page-001-102 | 52 | img-01516 | BHSv1S1 | Page-001-102 | 53 | img-01617 | BHSv1S1 | Page-001-102 | 57 | img-01718 | BHSv1S1 | Page-001-102 | 65 | img-01819 | BHSv1S1 | Page-001-102 | 70 | img-01920 | BHSv1S1 | Page-001-102 | 76 | img-02021 | BHSv1S1 | Page-001-102 | 80 | img-02122 | BHSv1S1 | Page-001-102 | 85 | img-02223 | BHSv1S1 | Page-001-102 | 88 | img-02324 | BHSv1S1 | Page-001-102 | 95 | img-02425 | BHSv1S1 | Page-001-102 | 100 | img-02526 | BHSv1S4 | Page-309-326 | 310 | img-02627 | BHSv1S4 | Page-309-326 | 311 | img-02728 | BHSv1S4 | Page-309-326 | 314 | img-02829 | BHSv1S4 | Page-309-326 | 315 | img-02930 | BHSv1S4 | Page-309-326 | 317 | img-03031 | BHSv1S4 | Page-309-326 | 320 | img-03132 | BHSv1S4 | Page-309-326 | 323 | img-03233 | BHSv1S4 | Page-309-326 | 324 | img-03334 | BHSv1S4 | Page-309-326 | 325 | img-03435 | BHSv1S4 | Page-309-326 | 326 | img-03536 | GG1 | page-002-100 | 3 | img-03637 | GG1 | page-002-100 | 4 | img-03738 | GG1 | page-002-100 | 5 | img-03839 | GG1 | page-002-100 | 6 | img-03940 | GG1 | page-002-100 | 13 | img-04041 | GG1 | page-002-100 | 14 | img-04142 | GG1 | page-002-100 | 15 | img-04243 | GG1 | page-002-100 | 17 | img-04344 | GG1 | page-002-100 | 19 | img-04445 | GG1 | page-002-100 | 24 | img-04546 | GG1 | page-002-100 | 26 | img-04647 | GG1 | page-002-100 | 28 | img-04748 | GG1 | page-002-100 | 62 | img-04849 | GG1 | page-002-100 | 65 | img-04950 | GG1 | page-002-100 | 74 | img-05051 | GG1 | page-002-100 | 78 | img-05152 | GG1 | page-002-100 | 79 | img-05253 | GG1 | page-002-100 | 80 | img-05354 | GG1 | page-002-100 | 81 | img-05455 | GG1 | page-002-100 | 82 | img-05556 | GG1 | page-002-100 | 83 | img-05657 | GG1 | page-002-100 | 84 | img-05758 | GG1 | page-002-100 | 85 | img-05859 | GG1 | page-002-100 | 92 | img-05960 | GG1 | page-002-100 | 100 | img-06061 | GG2 | page-101-200 | 101 | img-06162 | GG2 | page-101-200 | 102 | img-06263 | GG2 | page-101-200 | 103 | img-06364 | GG2 | page-101-200 | 104 | img-06465 | GG2 | page-101-200 | 106 | img-06566 | GG2 | page-101-200 | 108 | img-06667 | GG2 | page-101-200 | 110 | img-06768 | GG2 | page-101-200 | 113 | img-06869 | GG2 | page-101-200 | 115 | img-06970 | GG2 | page-101-200 | 121 | img-07071 | GG2 | page-101-200 | 125 | img-07172 | GG2 | page-101-200 | 127 | img-07273 | GG2 | page-101-200 | 128 | img-07374 | GG2 | page-101-200 | 129 | img-07475 | GG2 | page-101-200 | 131 | img-07576 | GG2 | page-101-200 | 134 | img-07677 | GG2 | page-101-200 | 139 | img-07778 | GG2 | page-101-200 | 148 | img-07879 | GG2 | page-101-200 | 156 | img-07980 | GG2 | page-101-200 | 170 | img-08081 | GG2 | page-101-200 | 172 | img-08182 | GG2 | page-101-200 | 189 | img-08283 | GG2 | page-101-200 | 195 | img-08384 | GG2 | page-101-200 | 196 | img-08485 | GG2 | page-101-200 | 199 | img-08586 | GG3 | page-201-300 | 201 | img-08687 | GG3 | page-201-300 | 204 | img-08788 | GG3 | page-201-300 | 205 | img-08889 | GG3 | page-201-300 | 206 | img-08990 | GG3 | page-201-300 | 214 | img-09091 | GG3 | page-201-300 | 217 | img-09192 | GG3 | page-201-300 | 220 | img-09293 | GG3 | page-201-300 | 222 | img-09394 | GG3 | page-201-300 | 224 | img-09495 | GG3 | page-201-300 | 234 | img-09596 | GG3 | page-201-300 | 240 | img-09697 | GG3 | page-201-300 | 251 | img-09798 | GG3 | page-201-300 | 254 | img-09899 | GG3 | page-201-300 | 267 | img-099100 | GG3 | page-201-300 | 269 | img-100101 | GG3 | page-201-300 | 280 | img-101102 | GG3 | page-201-300 | 289 | img-102103 | GG3 | page-201-300 | 296 | img-103104 | GG3 | page-201-300 | 298 | img-104105 | GG3 | page-201-300 | 300 | img-105106 | GOSv1S2 | page-101-150 | 101 | img-106107 | GOSv1S2 | page-101-150 | 102 | img-107108 | GOSv1S2 | page-101-150 | 106 | img-108109 | GOSv1S2 | page-101-150 | 107 | img-109110 | GOSv1S2 | page-101-150 | 108 | img-110111 | GOSv1S2 | page-101-150 | 116 | img-111112 | GOSv1S2 | page-101-150 | 118 | img-112113 | GOSv1S2 | page-101-150 | 142 | img-113114 | GOSv1S2 | page-101-150 | 143 | img-114115 | GOSv1S2 | page-101-150 | 144 | img-115116 | GOSv1S3 | page-151-195 | 151 | img-116117 | GOSv1S3 | page-151-195 | 154 | img-117118 | GOSv1S3 | page-151-195 | 158 | img-118119 | GOSv1S3 | page-151-195 | 166 | img-119120 | GOSv1S3 | page-151-195 | 168 | img-120121 | GOSv1S3 | page-151-195 | 170 | img-121122 | GOSv1S3 | page-151-195 | 172 | img-122123 | GOSv1S3 | page-151-195 | 175 | img-123124 | GOSv1S3 | page-151-195 | 185 | img-124125 | GOSv1S3 | page-151-195 | 194 | img-125

Fields of Study & Subjects

Handwritten Text Recognition Techniques Image Processing and 3D Reconstruction Digital Humanities and Scholarship

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
21 March 2026
Language
EN
Document type
Dataset