Fake-Face-Classification

Description

This repository hosts the source code, dataset, and resources for a cutting-edge project focused on detecting fake or manipulated faces using deep learning techniques. With the rise of digital manipulation tools, it has become increasingly important to develop robust methods for identifying fake images, especially in contexts such as social media and digital content verification.

In this classification task, I used the Regnet model with adabelief optimizer. Using pytorch utils for preprocessing and data splitting I used fake-face datasets that are generated by GAN models.

How to use:

First clone the repo

git clone https://github.com/Vargha-Kh/Fake-Face-Classification.git

Then install the requirements in the directory

pip install -r requirements.txt

Run main.py file to begin datasets preprocessing and start training

python main.py

Run evaluation.py to evaluate your trained model

python evaluation.py

Result

The final result of the classification evaluation was:
Accuracy: 1.0 Loss: 0.007738873939961195

Confusion matrix scores:
Precision: 100.0 Recall: 100.0, Accuracy: 100.0: ,f1_score: 100.0