Unmasking Deception: Deeplearning Approaches ...

Unmasking Deception: Deeplearning Approaches for Robust Deepfake Detection

Author Name : Sumit Yadav, Vishal Maurya, Vansh Jain, Piyush Sharma, Dr. Rekha Saha

ABSTRACT This review explores the progression in technology and detection methods for artificial media, in particular deepfake detection method, which are generated using deep learning algorithms. The development of effective deepfake detection models is essential due to the increasing preponderance and the outcome of deepfake content. In this work, we innovate "DeepFakeShield," a deep neural network (DNN) designed to accurately detect deepfakes. The model is based on the pre-trained VGG16 architecture and is evaluated on known deepfake datasets which comprises both real and fake images. "DeFakeShield" achieves an accuracy of 92.4% on the CelebDF-v2 dataset and 93.45% on the OpenForensics dataset. The model’s performance is compared against several state-of-the-art convolutional neural networks (CNNs) through extensive ablation studies. This research contributes to progressive deepfake detection techniques in managing artificial media in the arena of digital realm.