In online education scenarios, there is a “quasi-separation" between the instructors and the learners, making it difficult for instructors to perceive the emotional state of the learners. Therefore, studying the recognition of learners' emotions in online education can help instructors improve teaching strategies and also enable online education platforms to understand learners' learning preferences. At present, there have been many research achievements in the field of emotion recognition for online education learners. It's necessary to analyze and summarize from various perspectives. Firstly, the article elaborates on the model for representing emotions, which consists of three parts: the discrete model, the dimensional model, and the emotion categories of learners. Secondly, three methods for measuring emotions in online education and obtaining learners' emotional data are elaborated. Next, a summary of the methods for recognizing learners' emotions is provided, which includes text data, facial expressions, speech signals, physiological signals, and multimodal data. Finally, the article discusses the limitations and potential remedies in current research on the recognition of learners' emotions in online education. The article aims to conduct an in-depth analysis and summary of the related work on the emotional recognition of learners for online education, which provides valuable references for researchers in this field.