Deepfake-Classifier

A deep learning classifier to determine if an image is Real or Fake (Deepfake).

🎭 Deepfake Classifier

A project that applies Deep Learning to classify images as Real or Fake. Designed to help understand the real-world application of Convolutional Neural Networks (CNNs) in media forensics and misinformation detection.

---

📌 Project Goal

Use a trained Convolutional Neural Network (CNN) to:

• Classify user-uploaded images as "Real" or "Fake"
• Demonstrate how deep learning can be applied to detect digital forgeries
• Provide an open-source tool for learning, research, or further development in media verification

---

What this project does

This project is a binary image classifier for deepfake detection. It demonstrates how a pre-trained deep learning model can analyze an image to find subtle artifacts and patterns indicative of AI-generation.

It is not a comprehensive forensic analysis suite or a video detector. Instead, it focuses on a fundamental classification task:

“Is this image real or was it generated by AI?”

---

How it works

Given:

• A pre-trained model – e.g., EfficientNet, ResNet, or a custom CNN trained on a large dataset of real and fake images.
• Input Image – A user uploads an image file (e.g., JPG, PNG).

The classifier applies inference using the trained model to compute the output probability: the updated probability that the image is 'Fake' vs. 'Real'.

---

Why this matters

Deepfakes represent a significant challenge to digital trust. They can spread misinformation, commit fraud, and harass individuals. This tool makes the detection process explicit and transparent.

It can be useful as:

• An educational resource for students and data scientists learning about CNNs and computer vision
• A demo app for understanding the capabilities and limitations of deepfake detection
• A foundation to expand toward more robust, multi-modal (video/audio) detection systems

---

💡 What is a Convolutional Neural Network (CNN)?

A CNN is a class of deep neural networks, most commonly applied to analyzing visual imagery. They use a series of 'filters' (convolutions) to automatically learn and detect hierarchical features in an image, from simple edges to complex patterns like eyes or artifacts.

Simplified Flow: Input Image -> [Convolution + Pooling Layers] -> [Flattening] -> [Fully Connected Layers] -> Output (Real / Fake)

Where:

• Convolutional Layers: Scan the image to find patterns and features.
• Pooling Layers: Reduce the image size while retaining the most important information.
• Fully Connected Layers: Make the final classification decision based on the detected features.

---

🛠️ Features

• 🧠 Implements a deep learning model (e.g., EfficientNet, TensorFlow/Keras)
• 📤 Simple web interface to upload and classify images
• ⚙️ Simple, extensible Python script using Flask
• 👶 Beginner-friendly for open source contributors

---

🔍 Sample Use Case

Given:

• A user uploads an image of a person's face.
• The model analyzes the image for inconsistencies in lighting, shadows, skin texture, or eye reflections.

Output:

Classification: FAKE Confidence: 98.78%

---

🗂️ Project Struct

PROJECT/
├── .venv/
├── .vscode/
├── hardrvsf/
│   ├── fake/
│   └── real/
├── data.csv
├── logs/
├── rvsf/
│   ├── real_vs_fake / real-vs-fake/
│   │   ├── test/
│   │   ├── train/
│   │   └── valid/
│   ├── test.csv
│   ├── train.csv
│   └── valid.csv
└── python.ipynb

For a detailed, beginner-friendly explanation of what each file does, please read our guide:

➡️ View the Project Structure Guide

---

🚀 Getting Started

1. Clone the Repository

git clone [https://github.com/your-username/Deepfake-Classifier.git](https://github.com/your-username/Deepfake-Classifier.git)
cd Deepfake-Classifier

(Optional) Create and activate a virtual environment It's recommended to use a virtual environment to keep dependencies isolated.

On Windows (PowerShell):

python -m venv venv
.\venv\Scripts\Activate.ps1

On macOS / Linux:

python3 -m venv venv
source venv/bin/activate

2. Install Requirements

pip install -r requirements.txt

(Note: You may also need to download the pre-trained model file and place it in the /models directory if it's not included in the repo via Git LFS).

3. Run the App

python run.py

4. Open in Browser

[http://127.0.0.1:5000/](http://127.0.0.1:5000/)

Read the CONTRIBUTING.md file before making a pull request.

---

📜 License

This project is licensed under the MIT License.

---

🙌 Acknowledgements

This project was created and maintained by U Vikram and Alivia Hossain. Inspired by the challenge of combating digital misinformation and the fascinating field of generative AI.

---