We have collected 281 tongue crack images in JPG image format, and most of them are taken by experienced TCM doctors in standard environment. We use the object detection method to extract the bounding box of the whole tongue from the original image as the final dataset images. And we manually labeled them with Photoshop quick selection. Images in our dataset are with large structure variation for both foreground and background regions.
For some reasons, the dataset is not available at present.
If you have any academic questions or ideas, please feel free to communicate with us.
Jianqiang Peng, e-mail: 19210860025@fudan.edu.cn
Xinlei Li, e-mail: 18110860019@fudan.edu.cn