🤖 AI Summary
Existing fashion compatibility models struggle to incorporate users’ physical attributes (e.g., height, body shape) for personalized outfit recommendation. To address this, we propose the first fashion cognition learning paradigm grounded in individual physical characteristics and introduce an end-to-end Fashion Cognitive Network (FCN). FCN comprises two core components: a convolutional encoder that extracts visual-semantic embeddings of outfits, and a user-aware module that fuses appearance features. Additionally, we design a Multi-Label Graph Convolutional Network (ML-GCN) to jointly model fine-grained style semantics and fit compatibility. We curate a large-scale, real-world O4U dataset—comprising over 100K user-outfit interactions with annotated physical attributes and stylistic labels—to support training and evaluation. Extensive quantitative and qualitative experiments demonstrate that our method significantly outperforms state-of-the-art approaches in personalized outfit recommendation, achieving superior accuracy, interpretability, and generalization across diverse user demographics.
📝 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.
