An Improved Dense CNN Architecture for Deepfake Image Detection

Abstract

Recent advancements in computer vision processing need potent tools to create realistic deepfakes. A generative adversarial network (GAN) can fake the captured media streams, such as images, audio, and video, and make them visually fit other environments. So, the dissemination of fake media streams creates havoc in social communities and can destroy the reputation of a person or a community. Moreover, it manipulates public sentiments and opinions toward the person or community. Recent studies have suggested using the convolutional neural network (CNN) as an effective tool to detect deepfakes in the network. But, most techniques cannot capture the inter-frame dissimilarities of the collected media streams. Motivated by this, this paper presents a novel and improved deep-CNN (D-CNN) architecture for deepfake detection with reasonable accuracy and high generalizability. Images from multiple sources are captured to train the model, improving overall generalizability capabilities. The images are re-scaled and fed to the D-CNN model. A binary-cross entropy and Adam optimizer are utilized to improve the learning rate of the D-CNN model. We have considered seven different datasets from the reconstruction challenge with 5000 deepfake images and 10000 real images. The proposed model yields an accuracy of 98.33% in AttGAN a , 99.33% in GDWCT b , 95.33% in StyleGAN, 94.67% in StyleGAN2, and 99.17% in StarGAN c real and deepfake images, that indicates its viability in experimental setups. a Facial Attribute Editing by Only Changing What You Want (AttGAN) b Group-wise deep whitening-and-coloring transformation (GDWCT) c A GAN capable of learning mappings among multiple domains (StarGAN)

Unveiling Spoofing Attempts: A DCGAN-based Approach to Enhance Face Spoof Detection in Biometric Authentication

“…The structure of Arora et al [11], which focuses on the extraction and categorization of face characteristics, shows encouraging levels of accuracy but could run into problems with changing spoofing techniques and a variety of datasets. With potential for improvement, Patel et al [12] improved deep-CNN structure has great accuracy rates for a variety of false image types, including video deepfake detection.…”

“…When applied to diverse www.ijacsa.thesai.org datasets, the methodology may be exposed to modern spoofing approaches, and by enhancing and adapting the structure to address increasing threats and crimes, such as spoofing attacks on biometric systems. Patel et al [12] proposed an innovative and enhanced deep-CNN (D-CNN) structure for recognizing deep fakes that is both accurate and generalizable. Data from various sources are used for training the system, thereby boosting its overall generality characteristics.…”

Enhancing the Fight against Social Media Misinformation: An Ensemble Deep Learning Framework for Detecting Deepfakes

“…The extracted features are learned by the model during the training process; the network simultaneously analyzes the patterns in the image data. CNN further uses the learned expertise to determine which image is real or fake [35]. Moreover, CNNs are known for their ability to work with a large dataset which is very needful in image deepfake detection as voluminous samples are needed for the efficiency and effectiveness of the trained model.…”

Deepfake Generation and Detection: Case Study and Challenges

“…Deepfake comes from the underlying deep learning (DL) technology that swaps faces in digital content to create a fake impression of a person in a realistic environment. Deepfakes involve deep neural networks, convolutional neural networks (CNNs), autoencoders, and generative adversarial networks (GANs) as popular generation techniques [10] [11]. As deepfakes look realistic, it poses a great threat and questions the authenticity of the published content.…”