This is a Ai generated Fake Face and Real Face detection using Deepfake Machine Learning project  built with convolutional neural networks. The classifier was trained on a data set comprised of 1400 images (700 of each class) and tested on 600 images (300 per class). The classifier achieved an accuracy of **83.2%**. You can find more performance metrics and information about this project in
the repository. To use this web application just drag and drop a face image to be classified by the model. While you think about that, have a 🍪 and refresh the page once or twice to classify a few built-in faces embedded into the app. The classifier will return the result with the associated probability that a specific face image belongs to either the ```Real``` or ```Fake``` class. The model's architecture summary is also presented below

In recent years, advancements in artificial intelligence (AI) have led to the emergence of sophisticated techniques for generating fake images and videos, commonly known as deepfakes. These manipulations, facilitated by deep learning algorithms, have raised significant concerns regarding their potential to spread misinformation, manipulate public opinion, and infringe upon individuals' privacy and security.

Detecting deepfake content has become a crucial challenge in combating the proliferation of misleading information and protecting digital integrity. This project focuses on the development of a deep learning-based system for the detection of AI-generated fake faces, with the ultimate goal of distinguishing them from real faces.

The proliferation of deepfake technology has profound implications across various domains, including journalism, politics, entertainment, and cybersecurity. Misuse of deepfake content can lead to reputational damage, identity theft, and even exacerbate societal tensions. Therefore, developing robust techniques to identify deepfakes is imperative to mitigate these risks and safeguard the integrity of digital content.

This project aims to contribute to the ongoing efforts in deepfake detection by leveraging machine learning algorithms and computer vision techniques. By analyzing subtle discrepancies between real and fake faces, the proposed system seeks to provide a reliable means of identifying manipulated content and enhancing trust in digital media.

Technologies Used:

1. Deep Learning Frameworks:
   • TensorFlow or PyTorch: Widely-used frameworks for building and training deep learning models, including convolutional neural networks (CNNs) for image classification and detection tasks.
2. Computer Vision Libraries:
   • OpenCV: A popular library for computer vision tasks such as image preprocessing, feature extraction, and object detection.
   • scikit-image: Provides a collection of algorithms for image processing and manipulation, which can be useful for data preprocessing and augmentation.
3. Machine Learning Tools:
   • scikit-learn: Offers a range of machine learning algorithms and tools for data preprocessing, model evaluation, and metrics calculation.
   • XGBoost or LightGBM: Gradient boosting libraries that can be used for classification tasks, especially if ensemble methods are desired.
4. Streamlit and Web Development:
   • Streamlit: The primary framework for building interactive web applications with Python, allowing for the seamless integration of machine learning models with user-friendly interfaces.
   • Flask or FastAPI: Lightweight web frameworks that can be used for building backend APIs to support the Streamlit application.
5. Image Manipulation and Visualization:
   • Matplotlib or Seaborn: Libraries for creating visualizations and plots to display model outputs, evaluation metrics, and detection results.
   • Pillow: Python Imaging Library for opening, manipulating, and saving many different image file formats.
6. Other Utilities:
   • NumPy and Pandas: Fundamental libraries for numerical computing and data manipulation, which are essential for handling image data and preprocessing.
   • tqdm: Provides a progress bar for tracking the progress of data loading, model training, and inference tasks.

Installation

Use the package manager pip to install the requirements.txt file package.