Facial expressions are an important way for humans to perceive emotions. The advent of facial action coding systems has enabled the quantification of facial expressions. Moreover, a large amount of annotated data facilitates the performance of deep learning for the spotting and recognition of expressions or micro-expressions. However, the study of video-based expressions or micro-expressions requires coders to have expertise while also familiar with action unit (AU) coding. This paper systematically sorts out the relationship between facial muscles and AU to make more people understand AU coding from the principle. For this purpose, we have made a brief guide to get started as quickly as possible for the beginner to code.