Unveiling Text in Challenging Stone Inscriptions: A Character-Context-Aware Patching Strategy for Binarization Conference: ; ;
Pratyush Jena, Amal Joseph, Arnav Sharma, Ravi Kiran Sarvadevabhatla · 2026
Read Unveiling Text in Challenging Stone Inscriptions: A Character-Context-Aware Patching Strategy for Binarization Conference: ; ; by Pratyush Jena, Amal Joseph, Arnav Sharma, Ravi Kiran Sarvadevabhatla (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-03609.txt.
Description
Pratyush Jena and three co-authors work on reading ancient stone inscriptions in Indic scripts, where carved letters blend into weathered rock. The first step is binarization, turning a photo into clean black and white so letters stand out. Poor contrast, surface damage, stray marks, and wildly varying layouts make standard methods fail on these stones.
Their strategy cuts images into patches chosen adaptively, with character context guiding the choices, and trains an Attention U-Net network on them. Attention helps the model notice faint structural cues, while dynamic sampling teaches it to handle noise and irregular layouts. They also built a carefully annotated, pixel-precise dataset. Reliable binarization is the gateway to digitizing inscriptions before the stones erode further.
Catalog record
| CREATOR | Pratyush Jena, Amal Joseph, Arnav Sharma, Ravi Kiran Sarvadevabhatla |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 33 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-03609