Dress Well via Fashion Cognitive Learning
Kaicheng Pang, Xingxing Zou, Waikeung Wong
TL;DR
This work defines Fashion Cognitive Learning, a task that personalizes outfit recommendations by conditioning on individual physical attributes. It proposes the Fashion Cognitive Network (FCN), which combines an outfit encoder that aggregates attribute features via multi-size 1D convolutions with a two-layer Multi-label Graph Convolutional Network (ML-GCN) to capture label correlations among personal physical attributes. A new dataset, Outfits for You (O4U), containing 29,352 outfits and extensive personal features, supports evaluation and training. Empirical results show FCN outperforms strong baselines on multiple metrics, demonstrating improved precision in predicting which outfits are compatible with specific physical profiles, with potential for more accurate and trustworthy online fashion recommendations.
Abstract
Fashion compatibility models enable online retailers to easily obtain a large number of outfit compositions with good quality. However, effective fashion recommendation demands precise service for each customer with a deeper cognition of fashion. In this paper, we conduct the first study on fashion cognitive learning, which is fashion recommendations conditioned on personal physical information. To this end, we propose a Fashion Cognitive Network (FCN) to learn the relationships among visual-semantic embedding of outfit composition and appearance features of individuals. FCN contains two submodules, namely outfit encoder and Multi-label Graph Neural Network (ML-GCN). The outfit encoder uses a convolutional layer to encode an outfit into an outfit embedding. The latter module learns label classifiers via stacked GCN. We conducted extensive experiments on the newly collected O4U dataset, and the results provide strong qualitative and quantitative evidence that our framework outperforms alternative methods.