DeepBuzz: Deepfake Detection Tool

Live Link: https://huggingface.co/spaces/HarshitaSuri/DeepFake_Confidence_Score

DeepBuzz is an AI-powered solution designed to detect deepfake images with precision and confidence. The tool uses cutting-edge machine learning models, along with web scraping integrations, to analyze whether an uploaded image is real or fake, providing a detailed confidence score. The solution prioritizes privacy, accessibility, and usability for users looking to safeguard their digital identities.

Features

• Deepfake Detection:
  • Upload an image to determine whether it is real or a deepfake.
  • Powered by the Wvolf/ViT_Deepfake_Detection model from Hugging Face.
• Confidence Score:
  • The tool provides a confidence score to quantify its prediction.
• Web Scraping:
  • Utilizes SerpApi to search for images online based on keywords.
  • Integrates with ImgBB to securely upload and handle images.

Technologies Used

• Backend: Flask for server-side logic and API endpoints.
• Deepfake Detection Model: Wvolf/ViT_Deepfake_Detection (Hugging Face Transformers).
• Web Scraping Tools:
  • SerpApi: Fetches image results from search engines.
  • ImgBB: Handles uploading and hosting of images.
• Libraries:
  • transformers and torch for model inference.
  • Pillow for image processing.
  • requests for API communication.

Installation

Prerequisites

• Python 3.8 or higher
• API keys for:
  • SerpApi: Get your API key here.
  • ImgBB: Get your API key here.

Steps

1. Clone the repository: git clone https://github.com/TechVesrse-CT-University/CodeFusion