Answers are grounded in this paper and cite their sources.
The proliferation of AI-generated images, particularly in the domain of digital art, has raised significant challenges in distinguishing real content from synthetic creations. This research explores distinguishing between real and AI-generated images using art-painting datasets. A convolutional neural network (CNN)-based approach is proposed to classify images as either real or AI-generated, leveraging the CIFAR-10 network architecture for the task. The AI-generated paintings are produced using generative models such as GANs to simulate a wide range of artistic styles. Through rigorous training and evaluation, the results demonstrate the capacity and effectiveness of deep learning models in distinguishing between real and synthetic images. According to the comparing results of the CNN network and CIFAR-10 network, the accuracy of CIFAR-10 network is higher than the CNN network. The findings highlight model robustness in the detection of AI-generated content, providing insights into the future of automated content verification and digital forensics.
No takes yet. Share an insight, caveat, or question.
Oo et al. (2025) studied this question.