Rane7/CLP_dataset

Chinese landscape painting semantic segmentation dataset

《山水画语义感知解析数据集》（CLPD）是面向中国传统山水画场景理解与智能生成的结构化数据集。该数据集由华侨大学杨瑞博士团队及陕西师范大学吴晓军教授团队构建，旨在解决传统艺术图像数据来源分散、风格异构及语义模糊等难题。CLPD融合了故宫博物院等机构的数字文物资源，依据《芥子园画传》等画论确立了包含山、水、云、树等 14 类核心要素的语义标签体系，并引入人机协同标注与专家反馈机制，确保了标注的艺术准确性 。最终数据集包含 2206 幅 512×512 像素的高分辨率语义分割样本，覆盖宋、元、明、清四代主流画派。CLPD为山水画的数字化解析、风格聚类及生成式修复提供了坚实的数据基础 。

The Chinese Landscape Painting Semantic Perception Parsing Dataset (CLPD) is a structured dataset designed for scene understanding and intelligent generation in traditional Chinese landscape painting. Developed by the research teams of Dr. Rui Yang from Huaqiao University and Professor Xiaojun Wu from Shaanxi Normal University, this dataset addresses challenges such as fragmented data sources, stylistic heterogeneity, and semantic ambiguity in traditional art imagery. CLPD integrates digital cultural heritage resources from institutions like the Palace Museum. It establishes a semantic labeling system encompassing 14 core elements—including mountains, water, clouds, and trees—based on painting treatises such as The Mustard Seed Garden Manual of Painting. The dataset employs human-machine collaborative annotation and expert feedback mechanisms to ensure artistic accuracy in labeling. The final dataset comprises 2,206 high-resolution semantic segmentation samples at 512×512 pixels, covering mainstream painting schools from the Song, Yuan, Ming, and Qing dynasties. CLPD provides a robust data foundation for digital analysis, style clustering, and generative restoration of landscape paintings.

Dr. Rui Yang

Department of Artificial Intelligence

College of Computer Science and Technology

Huaqiao University

No.668 Jimei Avenue, Xiamen, Fujian, China 361021

🧐 Other Information

License: Please follow CC BY-NC 4.0.

Contact: Please contact Rui Yang by email.

📚 Citation

If you find the model, data, or code useful, please cite:

@article{10.1007/s00521-023-09343-w,
author = {Yang, Rui and Yang, Honghong and Zhao, Min and Jia, Ru and Wu, Xiaojun and Zhang, Yumei},
title = {Special perceptual parsing for Chinese landscape painting scene understanding: a semantic segmentation approach},
year = {2023},
issue_date = {Apr 2024},
publisher = {Springer-Verlag},
address = {Berlin, Heidelberg},
volume = {36},
number = {10},
issn = {0941-0643},
url = {https://doi.org/10.1007/s00521-023-09343-w},
doi = {10.1007/s00521-023-09343-w},
journal = {Neural Comput. Appl.},
month = dec,
pages = {5231–5249},
numpages = {19},
}