The ability of recent developments in deep learning generative models to produce incredibly realistic fake photos and movies has sparked worries. People's integrity is at risk, and social instability may result. New computational models that can effectively identify falsified information and notify viewers of possible picture and video modifications are desperately needed to address this problem. A thorough analysis of current research on deepfake content identification using deep learning-based techniques is presented in this study. By thoroughly examining the many types of false content identification, we hope to advance the state-of-the-art study. Additionally, we outline the benefits and limitations of the reviewed studies and suggest a number of future paths to address the problems and shortfalls in deepfake detection that remain unresolved.