Digital Image and Video Forgery Detection: A Comprehensive Review of Techniques, Datasets, and Future Directions
Dr. Pratibha V. Kashid
Associate Professor
Department of Information Technology
Sir Visvesvaraya Institute Of Technology Nashik
Ms.Gudulkar Gayatri Piraji1
Department of Information Technology
Sir Visvesvaraya Institute Of Technology Nashik
Ms.Katale Bhagyashree Shashikant2
Department of Information Technology
Sir Visvesvaraya Institute Of Technology Nashik
Ms. Avhad Anisha Balu3,
Department of Information Technology
Sir Visvesvaraya Institute Of Technology Nashik
Ms.Kurhade SakshiBhivaji4
Department of Information Technology
Sir Visvesvaraya Institute Of Technology Nashik
ABSTRACT
The rapid advancement of digital media editing tools and generative artificial intelligence has significantly increased the prevalence of manipulated images and videos. From traditional copy–move forgeries to sophisticated deepfake synthesis, digital content manipulation threatens information integrity, cybersecurity, journalism, and legal systems. This review provides a comprehensive analysis of digital image and video forgery detection techniques, benchmark datasets, evaluation metrics, and emerging research challenges. We categorize detection methods into classical forensic approaches and deep learning-based frameworks, examining their performance across detection accuracy, robustness, computational efficiency, interpretability, and multimodal integration. We further analyze major public datasets, highlight domain generalization challenges, and discuss adversarial vulnerabilities. Finally, we outline future research directions including cross-domain learning, physics-informed modeling, lightweight edge architectures, explainable AI, and self-supervised multimodal learning. This review serves as a unified technical reference for researchers, practitioners, and policymakers working in digital media forensics.
Keywords
Digital Forensics, Image Forgery Detection, Video Manipulation Detection, Deepfake Detection, CNN, Vision Transformers, Self-Supervised Learning
