International Journal of Advanced Innovative Technology in Engineering (IJAITE)

---

DeepFake Detection Using Inception-ResNet-v2

Dr. A. S. Manekar, Anand Wankhade, Aman Surkar, Pranjali Pande, Kunal Nemade

Abstract :

Control of pictures, recordings, and sounds utilizing face alter applications and web administrations have been being used, for quite a long time yet ongoing progress in deep learning have prompted AI-produced counterfeit pictures and recordings with traded faces, lip-adjusted sounds prominently known as Deepfakes. Deepfakes are created primarily using one of the following two methodologies: Autoencoders and Generator Adversarial Networks, both of which are based on pretrained deep neural networks. The level of authenticity accomplished by deep learning controlled deepfakes increments with expanding measures of information i.e., counterfeit pictures and recordings promptly accessible on the web at removal to prepare GANs. Deepfake algorithms make media leaving an uncovered edge of distinction between the true or unique source and the manufactured or deepfake objects. In this way, new systems and methods to distinguish through such deepfakes are the need of great importance. The methodology proposed based upon strong deep learning-based CNN designs to be specific, Inception-Resnet-V2 for recognizing the deepfakes. Our proposed approach not just surpasses the current methodologies regarding effectiveness and precision yet additionally offers the best concerning the given existence intricacy.

Keywords :

Full Text :

Download PDF

DOI :

Cite this paper :

References :

---