NORMA eResearch @NCI Library

DeepFake Detection using Deep Neural Networks

Agnihotri, Ambuj (2021) DeepFake Detection using Deep Neural Networks. Masters thesis, Dublin, National College of Ireland.

[thumbnail of Master of Science]
PDF (Master of Science)
Download (3MB) | Preview
[thumbnail of Configuration manual]
PDF (Configuration manual)
Download (3MB) | Preview


Deepfakes are fake images or videos created with artificial algorithms, image processing, and face swap. Deepfakes are computer-generated fake images or videos in which images are merged to create new images or videos representing events, comments, or activities that never occurred. The end product can be quite stunning. A ”Generative Adversarial Network,” or GAN, is an artificial intelligence technology that can be used to create fake images. GAN, a multifunction technique used to create Deep Fakes, is established to map faces using ”landmark” points. Such features include the edges of a person’s eyelids and mouth, nostrils, and the curve of the jawline. This research project’s main objective is to employ neural networks to distinguish between fraudulent and authentic images. For deepfake image detection, a publically available Flickr Faces High Quality (FFHQ) dataset is utilized. Deepfake image detection employs a variety of pre-trained Convolutional Neural Network (CNN) architectures (EfficientNetB4, InceptionV3, and InceptionResNetV2) for feature extraction and Long Short-Term Memory (LSTM) for classification. The Classification Report including Accucary, F-1 score, and other features are used to analyze the results. To execute code with essential python libraries such as Keras, Matplotlib, sklearn, and others, Google Colab and Jupiter Notebook are utilized. EfficientNetB4-LSTM, Inceptionv3-LSTM, and InceptionResNetv2-LSTM models achieved test accuracy of 98%, 96%, and 97%, respectively.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Deepfake; Generative Adversarial Network (GAN); Deep Learning; Convolutional Neural Network (CNN); Long Short-Term Memory (LSTM)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
Q Science > QA Mathematics > Computer software
T Technology > T Technology (General) > Information Technology > Computer software
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: Clara Chan
Date Deposited: 10 Nov 2021 10:20
Last Modified: 10 Nov 2021 11:43

Actions (login required)

View Item View Item