Deep-Fake Finder: Uncovering Forgery Image Through Neural Network Analysis

Digital images are a common form of media shared on social media, and their circulation can undermine news credibility and public trust. This research proposes a method for extracting image content, classifying it, verifying its authenticity, and identifying manipulations. With the exponential increase in social networking services, there has been a huge growth in the generation of image data, leading to the creation of fabricated images. These images are a major source of fake news, negatively impacting society. To verify the authenticity of these images, a method using a CNN deep learning model can be used. Pixel-based fake image detection can find tempering of an image up to a certain level. The raw image can be mangled in sections or as a whole, and it is necessary to recognize the type of image tampering and localize the tampered region. The image is converted into pixel format, and the RGB values are fed into the network's input layer. The output layer has two neurons for phony and genuine images, allowing us to infer if the image is phony and how likely it is to be tampered.