Personalized Fashion Recommendation Using Pairwise Attention

The e-commerce fashion industry is booming and comes with the need for proper search and recommendation. However, sufficient user personalization is still a challenging task. In this paper, we introduce a personalized fashion recommendation system based on high-dimensional input of user- and environment information. The proposed framework is used to estimate suitable categories and style of clothing depending on customized settings such as body type, age, occasion, or season. The goal is to recommend a full fitting outfit from the estimated suggestions. However, various personal attributes add up to a high dimensionality, and datasets are often very unbalanced or biased, making it difficult to do a proper recommendation. To solve this, we propose a pairwise-attention module to improve the performance of our framework. Our model can improve the performance up to 53.29% over the comparison method on MSE, mAP, and Recall. Moreover, in a subjective evaluation with human participants, the recommendations of the proposed method are preferred over the comparison method.