AI-Generated Sneaker Detection: Leveraging GANs and Convolutional Neural Networks for Image Classification

The increasing prevalence of AI-generated content presents unique challenges in the field of computer vision, especially when distinguishing between real and synthetic images. This study explores the detection of AI-generated sneakers, specifically from popular brands such as Nike, Adidas, and Converse, using Generative Adversarial Networks (GANs) and Convolutional Neural Networks (CNNs). The dataset for this project is a mix of real sneaker images sourced from Google Images and AI-generated images produced by the MidJourney AI platform. To enrich the dataset and enhance model training, synthetic images are generated through a GAN, providing a diverse range of examples. The primary objective is to train a robust detection model capable of distinguishing between real and AI-generated sneaker images by leveraging subtle visual differences. This research demonstrates the effectiveness of GANs in augmenting datasets for machine learning applications, while also testing the resilience of CNNs in distinguishing high-quality AI- generated images from authentic ones. The dataset, standardized to 240x240 pixel resolution, offers a comprehensive foundation for developing advanced image classification models aimed at tackling the growing challenge of AI-generated content detection.