Makeup transfer is not only to extract the makeup style of the reference\nimage, but also to render the makeup style to the semantic corresponding\nposition of the target image. However, most existing methods focus on the\nformer and ignore the latter, resulting in a failure to achieve desired\nresults. To solve the above problems, we propose a unified Symmetric\nSemantic-Aware Transformer (SSAT) network, which incorporates semantic\ncorrespondence learning to realize makeup transfer and removal simultaneously.\nIn SSAT, a novel Symmetric Semantic Corresponding Feature Transfer (SSCFT)\nmodule and a weakly supervised semantic loss are proposed to model and\nfacilitate the establishment of accurate semantic correspondence. In the\ngeneration process, the extracted makeup features are spatially distorted by\nSSCFT to achieve semantic alignment with the target image, then the distorted\nmakeup features are combined with unmodified makeup irrelevant features to\nproduce the final result. Experiments show that our method obtains more\nvisually accurate makeup transfer results, and user study in comparison with\nother state-of-the-art makeup transfer methods reflects the superiority of our\nmethod. Besides, we verify the robustness of the proposed method in the\ndifference of expression and pose, object occlusion scenes, and extend it to\nvideo makeup transfer. Code will be available at\nhttps://gitee.com/sunzhaoyang0304/ssat-msp.\n