Deep fake detection of videos using neural networks
Narikamalli Yaswanth, K. Kusuma, V. Durga Pujitha, Mohammad Basha, K. Mohan Kumar · 2025
The project endeavors to create a robust video classifier employing advanced deep learning architectures, including RESNET50, EfficientNetB2, InceptionV3, MobileNetV2, and VGG16. Initially, the system engages in comprehensive data preprocessing, accepting datasets comprising both original and morphed videos. Subsequently, it conducts feature extraction and deploys the selected classification model.Upon user input of a video, the system accurately discerns whether it is original or morphed, leveraging the chosen model&s;s capabilities. This endeavor is facilitated within the Google Colab development environment, harnessing TensorFlow and Keras libraries to streamline model development and evaluation.