Dynamic Style Generation of Clothing Based on Reinforcement Learning

This article aims to establish and validate a dynamic model for clothing style generation in order to enhance rapid style innovation and personalized customization in the clothing design sector. This study utilizes a clothing model grounded in CAD (Computer Aided Design) technology, paired with an RL (Reinforcement Learning) algorithm for style generation. By compiling and analyzing a comprehensive dataset of clothing CAD information and style reference samples, a simulation environment is created for model training and evaluation. The findings reveal that, in comparison to conventional CAD design techniques and rule-based style generation methods, the dynamic clothing style generation model presented in this study exhibits superior style consistency, originality, and aesthetic appeal. This model is capable of producing tailored clothing designs based on specified design elements and style references, demonstrating high levels of flexibility and adaptability. In conclusion, this research introduces an innovative design tool and model for the clothing industry, poised to streamline design processes, minimize costs, and foster sustainable industry growth. © 2024 U-turn Press LLC.