metadata
license: apache-2.0
Chinese Calligraphy Benchmark (ChineseCalligraphyBench, aka CCBench)
This dataset comprises of 120 famous Chinese calligraphy writings throughout ancient Chinese history, spanning from Western Zhou to Late Qing dynastry. It contains diversified writing styles including Oracle Bone Script, Zhuan (seal script), Li (clerical script), Kai (regular script), Cao (cursive script), and Xing (running script) etc. We give the traditional transcription as well as the corresponding simplified Chinese version.
The dataset could serve as an evaluation metric for OCR specialist models or generalist MLLMs.
Citation
@misc{boydcheung2026chinesecalligraphybench,
author = {Bo Zhang},
title = {ChineseCalligraphyBench},
year = {2026},
publisher = {Hugging Face},
url = {https://huggingface.co/boydcheung/ChineseCalligraphyBench}
}