Authors:
A. Kalki Prasad, D. Hemavathi

Addresses:
Department of Artificial Intelligence and Data Science, SRM Institute of Science and Technology, Kattankulathur, Chennai, Tamil Nadu, India. Department of Data Science and Business Systems, SRM Institute of Science and Technology, Kattankulathur, Chennai, Tamil Nadu, India.

Abstract:

Generative artificial intelligence has expanded swiftly, which has sparked a substantial rise in the number of deepfake video content that raises serious challenges to the authenticity and security of digital media. This paper introduces a web-based deepfake video detector that uses spatial feature analysis to detect AI-generated video manipulations. The presented method takes uploaded videos and extracts a set of representative images, which are then analysed by a convolutional neural network trained to detect real visual patterns and synthetic artefacts. Frame-level predictions are pooled to obtain an end-video authenticity score, enabling classification of videos as authentic or fake. To enhance transparency and interpretability, the system recognises and displays the key frames with a high probability of manipulation, and provides a visual explanation for the prediction. It includes secure user authentication and the continued storage of detection results to provide controlled access and traceability. As the experimental findings show, the proposed system offers effective and efficient deepfake detection, making it applicable in real-world practice.

Keywords: Deepfake Detection; Video Forgery; Computer Vision; Convolutional Neural Networks; Artificial Intelligence; Digital Media Forensics; Video Analysis; Spatial Feature Extraction; Multimedia Security; Authenticity Verification.

Received on: 09/04/2025, Revised on: 21/06/2025, Accepted on: 25/08/2025, Published on: 05/06/2026

DOI: 10.69888/FTSIN.2026.000705

FMDB Transactions on Sustainable Intelligent Networks, 2026 Vol. 3 No. 2, Pages: 75-84

• Views : 67
• Downloads : 8

Download PDF