Published Paper Details:

This paper presents a system based on Convolutional Neural Networks (CNNs) for detecting deepfake videos. It tackles the growing threat of AI-generated visual manipulation. Deepfake technology allows for the creation of very realistic videos that can spread false information, lead to identity theft, and harm reputations. The proposed system examines extracted video frames to find pixel-level inconsistencies and visual flaws that separate real content from fake. By using a lightweight, frame-based CNN architecture, the model achieves high accuracy and keeps computational demands low. This approach addresses the shortcomings of traditional deepfake detection methods. The system is trained on various datasets that include both authentic and manipulated videos. This helps improve its ability to generalize across different deepfake generation techniques. Experimental results show strong performance and scalability, confirming the model's ability to accurately detect deepfakes. This method provides a practical and efficient way to fight the misuse of deepfake technology in areas like cybersecurity, media verification, and digital forensics