Published Paper Details:

Deepfake detection is based on a deep learning model. Deepfake content is created with the help of artificial intelligence and machine learning to replace one person's face with another person's face in pictures. These manipulated pictures will undoubtedly have an enormous societal impact. Deepfake uses the latest technology like Machine Learning (ML), and Deep Learning (DL) to construct automated methods for creating fake content. We use the DFDC dataset for training the model, it contains 100,000 videos of more real and fake videos. To identify real and fake images using various Convolutional Neural Network (CNN) models, namely EfficientNetB0, VGG-16, and Xception. Our primary objective is to determine the model that yields the highest accuracy in identifying deepfake images. Initially, we preprocess videos from the DFDC dataset, extracting frames and preparing them for input into CNN models. Subsequently, we construct and train each CNN architecture on the preprocessed data, integrating techniques like data augmentation to enhance model generalization and resilience. Performance evaluation utilizes metrics such as accuracy, precision, recall, and F1 score.