Auricular diagnosis, as an important part of traditional medicine, is easy preformed and non-invasiveness. The specific regions of auricular area can reflect health conditions of the corresponding organ systems. Literature review indicates that deep learning studies in the auricular regions segmentation have been limited compared with facial and tongue deep learning studies in traditional Chinese medicine (TCM) due to the lack of annotated datasets. In this study, two novel auricular datasets are constructed. One is AMI Ear dataset with key-organ mapping region annotations, the other is WYEar dataset. Each auricular image is labelled with five specific mapping regions: heart, liver, spleen, lung, and kidney. These two datasets can be applied for the segmentation of TCM key-organ mapping regions in auricular image.