DEEP FAKE VIDEO DETECTION USING CONVOLUTIONAL NEURAL NETWORKS
DOI:
https://doi.org/10.62643/Abstract
With rapid advancements in artificial intelligence, highly sophisticated techniques are now being used to generate realistic fake videos.
These manipulated media forms, known as deepfakes, pose serious risks to society, particularly in social and political contexts, where they
can be exploited for harmful intentions. Deepfakes are digitally altered videos or representations created using advanced AI methods,
producing fabricated visuals and audio that closely resemble real individuals. By analyzing images and videos of a person from multiple
perspectives, deep learning systems can imitate their facial expressions, movements, and speech patterns with high accuracy. The
detection of such content has become increasingly challenging due to continuous improvements in deepfake generation technologies,
making the fabricated media more convincing and difficult to identify. Addressing this issue, the proposed work focuses on developing a
model that examines video frames using deep learning techniques to identify irregularities in facial structures, compression artifacts, and
other subtle inconsistencies introduced during the creation process.The model incorporates a Convolutional Neural Network (CNN)
combined with transfer learning to effectively recognize these hidden anomalies. It is trained to detect the distortions and artifacts
commonly present in deepfake videos, especially around facial regions. For training and evaluation, the model utilizes the “Celeb-DF: A
New Dataset for DeepFake Forensics,” which provides high-quality samples for analysis.
Downloads
Published
Issue
Section
License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.













