The technological developments in deep learning have led to the development of very realistic face swap deepfake videos, which are a significant threat to digital security and trust. Deepfake videos, developed using methods such as Generative Adversarial Networks (GANs), autoencoders, and neural face swap algorithms, have the capability to manipulate facial identities in a very realistic way, making it extremely difficult to manually identify them. The objective of this research work is to develop a comprehensive AI/ML system for the identification of face swap deepfake videos using spatial and temporal facial feature analysis. The proposed system combines frame extraction, facial landmark detection, temporal inconsistency modeling, and a custom-developed Convolutional Neural Network (CNN) model for binary classification of videos into REAL and FAKE categories. The latest preprocessing methods such as facial region cropping, normalization, and data augmentation are used to improve the robustness of the model and prevent overfitting. Temporal feature aggregation is also employed to detect the unnatural blending artifacts, irregular blinking rates, and frame distortions that are generally present in manipulated videos. The experimental evaluation is conducted by utilizing structured training, validation, and testing data. The proposed model performs well on the testing data with a high accuracy of 91.47%, very low Binary Cross-Entropy loss, and an excellent Area Under the Curve (AUC) value of 0.92, which shows a strong classification ability. The performance comparison of the proposed model with the existing CNN and hybrid models confirms the superiority of the proposed model in generalization performance and detection robustness. The experimental result clearly shows that the combination of spatial and temporal feature analysis is an important factor in enhancing the effectiveness of deepfake detection. Future work includes the design of transformer models, real-time systems, and adversarial robustness enhancement.
Deepfake detection; Deep learning; Customize CNN; Deepfake Detection Challenge Dataset; Classification.
International Journal of Trend in Scientific Research and Development - IJTSRD having
online ISSN 2456-6470. IJTSRD is a leading Open Access, Peer-Reviewed International
Journal which provides rapid publication of your research articles and aims to promote
the theory and practice along with knowledge sharing between researchers, developers,
engineers, students, and practitioners working in and around the world in many areas
like Sciences, Technology, Innovation, Engineering, Agriculture, Management and
many more and it is recommended by all Universities, review articles and short communications
in all subjects. IJTSRD running an International Journal who are proving quality
publication of peer reviewed and refereed international journals from diverse fields
that emphasizes new research, development and their applications. IJTSRD provides
an online access to exchange your research work, technical notes & surveying results
among professionals throughout the world in e-journals. IJTSRD is a fastest growing
and dynamic professional organization. The aim of this organization is to provide
access not only to world class research resources, but through its professionals
aim to bring in a significant transformation in the real of open access journals
and online publishing.