Year: 2024
Type: conference-paper
Abstract: The advent of deepfake technology has increased various cybersecurity vulnerabilities.Mitigating these threats poses a significant challenge due to the realistic nature and difficulty of detecting deepfakes.The rise of AI-driven tools introduces additional intricacy and intrigue to generating exceptionally realistic deepfake images.The photos generated by these technologies exhibit a remarkable level of realism, making it extremely difficult for human observers to differentiate between deepfake and real images.To better understand recent advances, we explored state-of-the-art techniques in image generation.Our investigation reveals a gap in the existing literature: There are no datasets of fake faces in the Indian context.To address this, we develop a web application capable of generating high-quality fake faces and creating a dataset of Indian fake faces.In addition, we develop an Android application specifically for deepfake detection tasks.
Authors: Surbhi Raj, Jimson Matthew, Arijit Mondal
Institution: Indian Institute of Technology Patna
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Award: G866603823
OpenAlex ID: W4412638192