Combating Image Deception: A Convolutional Neural Network Approach

ABSTRACT

A ACsort of synthetic media known as "deepfakes" occurs when someone's likeness is subtly substituted into already-existing content, with the intention of misleading the public. Deepfake photos have become more and more common in recent years, partly due to the easy-to-use, open-source programs such as FakeApp that makes it easy to generate deepfakes images and further shared on social media. A number of societal issues, such as disinformation, political choas, and the damage of public persons' reputations, have been sparked by this surge. In response, a deepfake detection model trained on a 140,000 images dataset publically available on Kaggle is presented in this research. Our model, which makes use of Convolutional Neural Network (CNN) architecture, achieves a remarkable accuracy of 95.25%, providing a viable tactic in the ongoing fight against visual fraud.

Keywords: CNN, Deepfakes, Deeplearning, Social media.