Published in: Journal of Computer Science and Engineering in Innovations and Research (ISSN No: 3049-1762 online)
Publication Date: May 15, 2025
Deepfake technology uses deep learning and AI to create highly realistic yet fabricated images and videos by manipulating a person’s likeness. While it can be used for creative purposes, deepfakes pose significant risks, including misinformation, identity fraud, and public manipulation. Detecting deepfakes involves identifying subtle visual inconsistencies, such as unnatural facial features, blinking, or lighting, but as the technology advances, these errors become harder to spot. Detection methods, including AI-based models, must continually evolve to keep pace with new deepfake creation techniques. Ethical concerns also arise as deepfake technology becomes more accessible, making it easier for malicious actors to produce harmful content. To address these challenges, detection systems need to incorporate advanced techniques, such as analyzing audio, metadata, and context, to preserve the integrity of digital content and protect against misuse. In this study, a technique has been presented using a fine-tuned DenseNet201, a convolutional neural network model, to detect deepfake pictures from video inputs. The Adam optimizer is used in this study to calculate and adjust the required parameters.
Deep Learning Techniques, Deep Fake Video, DenseNet Architecture