Designers are always pursuing design with suitable emotions. Effective emotional fusion not only produces a good user experience but also extends the product lifecycle. The decoding of design emotion and the use of design emotion language should run through the entire design process. In this study, we propose a new emotion-embedded design flow (EFlow) based on design big data and deep learning technology. This method focuses on how emotion is input into the design process and improves the effectiveness of emotional design. An emotion database containing 2054 labeled images is collected and a deep fuzzy classification network is proposed. Through realizing the automatic emotional judgment of the design reference materials and the design output content using the deep learning technology, EFlow not only saves manpower and test cost but also provides a reference that a designer can use to optimize and improve the design process. It promotes a new way of thinking about connecting artificial intelligence technology and the design field.