TCM Facial Diagnosis (TCM-FD) dataset

Traditional Chinese Medicine (TCM) facial inspection offers a non-invasive approach to health assessment, but its reliance on subjective practitioner judgment hinders standardization and widespread application. To address this challenge and facilitate the development of objective, AI-driven diagnostic tools, we present the TCM-FD (Traditional Chinese Medicine Facial Diagnosis) Dataset. This dataset contains 20,182 high-resolution facial images, curated from the publicly available Flickr-Faces-HQ (FFHQ) dataset.The core contribution of this dataset lies in its meticulous, multi-task annotations. Each image has been labeled for 11 distinct diagnostic indicators\u2014such as Overall Complexion, Lip Color, and Scleral Color\u2014by a panel of eleven senior TCM practitioners. The final ground-truth labels were established through a majority vote, ensuring a high degree of clinical relevance and reliability. The data is structured with individual .png image files and corresponding .xml annotation files, providing a clear and accessible format for machine learning applications.The TCM-FD dataset provides an essential resource for the biomedical informatics and computer vision communities to advance research in automated medical diagnosis, multi-task learning, and the objective standardization of traditional medicine.

搜集汇总

背景与挑战

背景概述

该数据集包含20,182张高分辨率面部图像，由11位资深中医师对11个诊断指标进行多任务标注，旨在推动基于AI的客观中医面部诊断研究，为生物医学信息学和计算机视觉领域提供标准化资源。