Design and application of an intelligent generation model for fashion clothing images based on improved generative adversarial networks

Achieving intelligent fashion pattern design has always been a challenging task in the field of fashion design. Traditional manual design methods require a significant amount of time and manpower, and have limitations in understanding and expressing various patterns. To overcome these issues, research has improved the standard deep convolutional generative adversarial network (DCGAN) by introducing new network structures and loss functions. Real time style conversion techniques have been combined with the improved DCGAN to achieve fast and efficient clothing pattern generation. Finally, the standardization technique of StyleGAN has been introduced to further enrich the diversity and details of the generated images. As a result, a novel intelligent fashion image generation model is designed. Experimental results show that the model has low CPU usage, with an average usage rate of approximately 10.8% and a peak usage rate of approximately 71.5%. In the ImageNet dataset, the adversarial loss function converges to 0.35 after 370 iterations. The fashion image generation model developed in this study has higher clarity and lower resource consumption, demonstrating potential applications in the field of image generation.