Research on the semantic segmentation of Thangka images via an improved PIDNet
Jiao Wu, Tiejun Wang, Xiaoyan Hu, LingMei Tao, Tianjiao Duan, Yanjiao Wei · 2026
Read Research on the semantic segmentation of Thangka images via an improved PIDNet by Jiao Wu, Tiejun Wang, Xiaoyan Hu, LingMei Tao, Tianjiao Duan, Yanjiao Wei (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/plos-pone-0332164.txt.
Description
Jiao Wu and colleagues improved a fast image analysis model so it can outline the figures in Thangka paintings, the intricate sacred scrolls of Tibetan Buddhism. In these images, deities blend into busy backgrounds filled with ritual objects and ornament, blurring the edges that computer vision needs. Their upgrades sharpen attention on informative details and blend local and global features while keeping the model quick.
On a custom Thangka dataset the model labelled regions with solid accuracy while running at more than one hundred frames per second, fast enough for live use. It also generalised well on a standard city-scene benchmark, beating its baseline there too. The authors offer the tool for digital preservation and restoration work, where scholars must map damaged or crowded paintings quickly and reliably.
Catalog record
| CREATOR | Jiao Wu, Tiejun Wang, Xiaoyan Hu, LingMei Tao, Tianjiao Duan, Yanjiao Wei |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 57786 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |