📰 Fake News Detection System (Multi-Modal)

An advanced AI-powered web application that detects fake news across multiple formats — including text articles, AI-generated images, and deepfake videos. Built with modern ML/NLP/CV techniques, it helps reduce misinformation by analyzing and classifying the authenticity of various content types in real time.

---

🚀 Features

• 🧠 Detects fake vs real news articles using NLP + ML
• 🖼️ Detects AI-generated images using CNNs and Transformer models
• 🎞️ Detects deepfake or tampered videos using video forensics techniques
• 📑 Text classification using pretrained transformer models (e.g., BERT)
• 🌐 Web-based UI built with React for easy testing
• ⚡ FastAPI backend with endpoints for text, image, and video input
• 🧾 Clean and simple UI for testing all media types
• 🔁 Trained on diverse datasets (Kaggle, FakeNewsNet, FF++)

---

🛠️ Tech Stack

Frontend:

• React.js
• Axios

Backend:

• Python
• FastAPI
• Scikit-learn
• HuggingFace Transformers
• OpenCV
• TensorFlow / PyTorch (for CV models)
• NLTK, NumPy, Pandas

---

📦 Installation & Setup

🔧 Prerequisites

• Python 3.8+
• Node.js & npm
• pip or virtualenv
• CUDA (optional for GPU-based video/image models)

---

1️⃣ Clone the repository

git clone https://github.com/your-username/fake-news-detector.git
cd fake-news-detector

2️⃣ Backend Setup (FastAPI)

cd backend
pip install -r requirements.txt
uvicorn main:app --reload

📍 The FastAPI backend will start at: http://localhost:8000

3️⃣ Frontend Setup (React)

cd frontend
npm install
npm start

📍 The React app will start at: http://localhost:3000

📥 Input Types Supported

📝 News articles (text)

🖼️ Image files (e.g., .jpg, .png)

🎞️ Video files (e.g., .mp4, .avi)

Each content type is analyzed separately by a dedicated ML model.

📈 How It Works

📝 Text: User enters news content → Sent to backend → NLP model analyzes → Returns ✅ Real or ❌ Fake.

🖼️ Image: User uploads image → CNN/Transformer model checks for AI generation → Returns result.

🎞️ Video: User uploads video → Frames analyzed for deepfakes → Returns authenticity status.

Each result is shown instantly on the frontend with a confidence score.

📈 Output

✅ Real / ❌ Fake classification

⚠️ Confidence scores

📊 Optionally show visual explanation (Grad-CAM, attention maps)

📸 Frontend Screenshots

Screenshots of the user interface for testing news, images, and videos.

📝 News Detection UI