Deepfake Detection App

A web application for detecting deepfakes in images, videos, and audio using CNN and Wav2Vec2 models.

Models

Media Type	Model	Architecture
Image	deepfake_cnn.keras (included)	CNN with 4 Conv2D blocks
Video	Uses image model	Frame-by-frame analysis
Audio	Wav2Vec2	Transformer-based

Setup

1. Install dependencies:

pip3 install -r requirements.txt

2. Download the audio model (optional, for audio detection):

git lfs install
git clone https://huggingface.co/mo-thecreator/Deepfake-audio-detection

3. Run the application:

python3 app.py

4. Open your browser: http://127.0.0.1:5001

Features

• Image Analysis: Analyzes images for deepfake artifacts using CNN
• Video Analysis: Extracts frames and analyzes each with the image model
• Audio Analysis: Uses Wav2Vec2 transformer model for voice deepfake detection

Supported Formats

• Images: PNG, JPG, JPEG, GIF, BMP, WebP
• Videos: MP4, AVI, MOV, MKV, WebM
• Audio: MP3, WAV, OGG, FLAC, M4A

Chrome Extension

1. Open Chrome → chrome://extensions/
2. Enable Developer mode
3. Click Load unpacked → Select the extension folder
4. Run the Flask app (python3 app.py) — it listens on http://127.0.0.1:5001 (port 5001 avoids macOS AirPlay using port 5000)
5. Toolbar: Open the popup to upload files
6. Right‑click any image on a webpage → Analyze image for deepfake — opens a full tab so the window stays open (the small popup closes if you drag files elsewhere; use this for images on the page)

The extension needs permission to reach your local server and to download images you right‑click (http/https).

API

POST /analyze

curl -X POST -F "file=@image.jpg" -F "type=image" http://127.0.0.1:5001/analyze

Response:

{
  "is_fake": true,
  "confidence": 87.5,
  "fake_probability": 87.5,
  "real_probability": 12.5,
  "message": "FAKE detected with 87.5% confidence"
}

Project Structure

ml_Cp/
├── app.py                    # Flask backend with ML inference
├── requirements.txt          # Python dependencies
├── deepfake_cnn.keras/       # Image deepfake CNN model (included)
├── Deepfake-audio-detection/ # Audio model (download separately)
├── templates/
│   └── index.html            # Web UI
├── extension/                # Chrome extension
└── uploads/                  # Uploaded files (auto-created)

Requirements

• Python 3.9+
• TensorFlow 2.16+
• PyTorch 2.1+
• ~1GB disk space for models