Deepfake Detection In Multi-Model Using Deep Learning

Abstract

Artificial Intelligence (AI) refers to the broad field where machines are trained to carry out tasks such as learning, reasoning and decision-making. Deep Learning (DL) is a branch of AI that relies on multi-layered neural networks to extract features automatically, handling complex data like images and videos with high accuracy, which makes it well-suited for deepfake detection. Current detection systems mostly depend on Convolutional Neural Networks (CNNs) to examine spatial artifacts in video frames, alongside frequency-domain analysis to pick up subtle inconsistencies. While these approaches yield solid results on benchmark datasets and underpin much of the existing detection research, they tend to fall short when dealing with low-quality, compressed or rapidly changing fake videos. To address these limitations, we introduce a multi-modal deep learning framework that combines CNNbased video preprocessing, LSTM temporal analysis and audio feature extraction with ongoing retraining. Together, these elements considerably improve the robustness, adaptability and accuracy of deepfake detection in practical settings, reaching 98.47% accuracy on the FaceForensics++ benchmark