Detecting AI-Generated Artworks With Deep Learning

The rapid development of artificial intelligence has made an increase in the amount of generated content around the globe. Art is the most threatened field due to AI, potentially affecting the market value of an artist's creation. These generated images also affect artists who rely on their work for income as the accessibility of these models diminishes the demand for hiring real artists. As these generative models continue to evolve, it is becoming more difficult to differentiate between authentic art and generated ones. This paper addresses the need for a tool that can accurately distinguish between human-made and AI-generated artworks using a convolutional neural network (CNN) known as EfficientNetV2, ResNet50, and VGGNet. Our methodology involves using a dataset comprising images created by both human artists and AI algorithms. This aim of developing this tool is to preserve the integrity of human-made art while also protecting artists' rights and sustaining the value of art in the market. The researched produce a satisfying result with EfficientNetV2 achieving 98.7% accuracy, ResNet50 achieving 98.5% accuracy and VGG16 achieves 97.4% accuracy. This solves our problem of identifying AI-generated artworks as artists can rest assure their integrity and originality of their work can be recognize by using this model.