Published Paper Details:

The rapid advancement of deep learning technologies has led to a surge in deepfake media--synthetically generated images and videos that closely mimic real human appearances and expressions. These forgeries pose a significant threat to digital security, privacy, and the credibility of online content. This paper explores a hybrid deep learning framework that integrates Convolutional Neural Networks (CNNs) for feature extraction with Support Vector Machines (SVMs) for classification to enhance the accuracy and robustness of deepfake detection. CNNs are leveraged to automatically learn and extract high-dimensional spatial features from facial images and video frames, while SVMs provide a powerful mechanism for discriminating between real and manipulated media based on the extracted features. Experimental results across multiple deepfake datasets demonstrate that the hybrid CNN-SVM model outperforms traditional standalone models in terms of precision, recall, and F1-score, showcasing its effectiveness for real-world deployment. The proposed approach offers a promising direction for combating the proliferation of deepfakes through intelligent fusion of deep and classical machine learning techniques.