Tencent Hunyuan Open-Sources Chronicles-OCR Benchmark for Ancient Chinese Characters
AI HOT 精选157 字 (约 1 分钟)
75
Tencent Hunyuan has open-sourced Chronicles-OCR, a visual perception benchmark evaluating VLLMs on ancient Chinese characters. The dataset spans 3,000 years of evolution, covering 7 historical scripts from Oracle Bone to Cursive with 2,800 balanced images, and includes 4 core tasks: Character Spotting, Fine-grained Recognition, Ancient Text Parsing, and Script Classification.
入选理由：数据集覆盖3000年汉字演变，包含7种历史字体和2800张平衡图像
FeaturedArticle#OCR#VLLMs#Ancient Chinese Characters#Benchmark#Tencent Hunyuan英文