CRGAN: A Context-Aware Clothing Design and Recommendation System for Young Sri Lankan Females Using Generative Adversarial Networks

The fashion industry is constantly undergoing notable transformations fueled by advancements in technology, particularly AI and ML. This shift is driven by the desire for personalized clothing recommendations, especially among young females, who seek tailored suggestions based on their preferences, climate, and style. While traditional recommendation systems rely on existing databases to suggest predefined outfit options, this research takes a novel approach by integrating generative modeling techniques to create unique outfit designs based on contextual factors. The proposed system leverages a Conditional Generative Adversarial Network (GAN) model, trained on three predefined parameters—attire type, temperature conditions, and the user’s skin tone. This innovative integration of context-aware features marks a significant advancement over prior research, which has largely neglected these combined parameters, particularly in the context of cultural demographics. Targeting young Sri Lankan females, this study fills a critical gap in generative clothing systems by addressing the unique cultural and environmental needs of this population. Experiment results on a labeled dataset demonstrate the model’s ability to generate accurate and personalized outputs based on up to three input parameters. Results are verified through quantitative model evaluations and user studies, underscoring the system’s potential to redefine personalized fashion recommendations.