Ram Kumar Sharma – Assistant Professor, Dept. of IT, NIET, Greater Noida, India. ramsharma533@gmail.com
Abstract
Deepfake technology leverages **Generative Adversarial Networks (GANs)** to create **synthetic media**, making detection a critical challenge in **digital forensics**. This study presents a **CNN-based detection model**, trained on publicly available datasets, capable of identifying **spatial inconsistencies and subtle artifacts** present in deepfake-generated content.
The **CNN model developed for deepfake detection** exhibited **strong performance**, successfully identifying **synthetic artifacts left by generative networks**. Future improvements include **enhanced image quality sensitivity, temporal context integration, and real-time deployment**.
References
Afchar et al. (2018). “MesoNet: a Compact Facial Video Forgery Detection Network.”
Rossler et al. (2019). “FaceForensics++: Learning to Detect Manipulated Facial Images.”
Dolhansky et al. (2020). “The Deepfake Detection Challenge Dataset.” arXiv.
2 comments