DEEPFAKE FACE DETECTION OF IMAGES USING TRANSFER

Abstract

With the rapid advancement of artificial intelligence and deep learning technologies,
the creation of highly realistic manipulated media, commonly known as deepfakes,
has become increasingly widespread. Deepfake images are generated using
sophisticated algorithms that can alter or replace facial features, making it difficult to
distinguish between real and fake content. This raises serious concerns related to
misinformation, identity theft, and digital security, making deepfake detection a
critical problem in computer vision and digital forensics.
This project presents an effective approach for detecting deepfake face images using
transfer learning techniques. The system utilizes a pre-trained deep learning model,
MobileNetV2, to extract high-level features from facial images and classify them as
real or fake. By leveraging transfer learning, the model benefits from knowledge
gained from large-scale datasets, resulting in improved accuracy and reduced training
time.
The dataset used consists of both real and manipulated facial images obtained from
publicly available sources. Preprocessing steps such as image resizing and
normalization are applied to ensure data consistency. The dataset is then divided into
training, validation, and testing sets for proper evaluation of the model. During
training, MobileNetV2 acts as a feature extractor, while additional layers are used for
binary classification. The model learns to identify subtle inconsistencies in facial
structures, textures, and patterns that indicate manipulation.