Research on the semantic segmentation of Thangka images via an improved PIDNet.
Clinical Snapshot
PICO Framework
| P — Population | Thangka paintings (Tibetan Buddhist art) requiring digital preservation and analysis |
| I — Intervention | Improved PIDNet model with ECA-Pag and LGFM modules plus PConv3 |
| C — Comparator | Conventional segmentation methods and baseline PIDNet |
| O — Outcomes | Semantic segmentation accuracy (mIoU), boundary detection (mB-Fscore), processing speed (FPS) |
Bottom Line
This technical study presents an improved computer vision method for automatically segmenting Thangka paintings, achieving 73.28% accuracy while maintaining real-time processing speeds. While the methodology appears sound and addresses a genuine need in cultural heritage preservation, the study lacks statistical rigor with no confidence intervals or significance testing reported. The practical application is limited to specialised cultural preservation contexts, though the underlying technology may have broader computer vision applications. For Australian institutions managing Asian art collections or digital humanities projects, this represents a potentially useful tool, though implementation would require significant technical expertise. The study would benefit from more robust statistical analysis and clearer methodology reporting to strengthen its evidence base.
Key Findings
P Value: Not reported
Effect Size: 73.28% mIoU achieved on custom Thangka dataset
Primary Outcome: Semantic segmentation accuracy measured by mean Intersection over Union (mIoU)
Nnt Or Sensitivity: mB-Fscore of 40.01% for boundary detection accuracy
Confidence Interval: Not reported
Clinical Application
Requires specialised computer vision expertise and appropriate computational hardware Relevant for Australian museums with Asian art collections, universities with digital humanities programs, and cultural institutions collaborating with Tibetan communities Cultural heritage institutions, museums, and digital preservation specialists working with Tibetan Buddhist art
Abstract
In the digital preservation, restoration, and research of Thangka paintings, real-time semantic segmentation plays a crucial role in rapid image analysis. However, Thangka images exhibit intricate compositions, where principal figures often blend with backgrounds, ritual objects, and intricate ornaments, leading to blurred boundaries and fine details that challenge conventional segmentation methods in balancing accuracy and efficiency. To address this, we propose an improved PIDNet-based model incorporating ECA-Pag (Efficient Channel Attention-Path Aggregation) and LGFM (Local-Global Feature Fusion Module) modules, along with PConv3 (Partial Convolution 3), enhancing feature extraction and segmentation precision. Experimental results demonstrate that our model achieves 73.28% mIoU and mB-Fscore (mean Boundary F-score) of 40.01% on a custom Thangka dataset while maintaining 109.03 FPS (Frames Per Second), ensuring both high accuracy and real-time performance. Furthermore, evaluations on the Cityscapes benchmark confirm the model's generalization capability, outperforming baseline methods. This work provides an efficient and reliable solution for Thangka image segmentation, with potential applications in cultural heritage preservation and broader computer vision tasks.
References
- 1.Wu, J., Wang, T., Hu, X., Tao, L., Duan, T., & Wei, Y. (2026). Research on the semantic segmentation of Thangka images via an improved PIDNet. PLoS One. https://doi.org/10.1109/iccv48922.2021.00986
Related Research
PloS one
Empowering rural governance with digital technology: Deep learning models for automated detection of rural buildings using remote sensing images
14 June 2026
JMIR medical education
Performance of Large Language Models on the Brazilian National Medical Education Examination: Comparative Benchmark Study
31 May 2026
PloS one
Transformation of artistic style and innovative design of oriental folk patterns based on AIGC Technology-A case study of Zhuxian town new year paintings from China
30 May 2026
This content is for educational purposes for healthcare professionals only and does not constitute clinical advice. Clinical decisions should be based on individual patient assessment, current guidelines, and appropriate specialist consultation. Editorial Standards · Privacy Policy · Terms of Service