🛡️ ShivaSpear Deepfake Detection

🚀 Overview

ShivaSpear Deepfake Detection is an AI-powered tool for identifying deepfake images, audio, and fake news. It leverages machine learning and natural language processing (NLP) to combat misinformation.

🔥 Key Features

✅ Deepfake Image & Audio Detection – High-accuracy identification of manipulated media.
✅ Fake News Classification – NLP-powered fake news detection.
✅ Preprocessing Tools – OpenCV, Librosa, and NLP libraries for data refinement.
✅ User-Friendly UI – Streamlit-powered web app for real-time analysis.
✅ Fast & Efficient – Optimized models for quick inference.

🛠️ Tech Stack

Category	Technologies
ML & AI	TensorFlow, Keras, PyTorch (optional), Scikit-learn
Image Processing	OpenCV, Pillow (PIL)
Audio Processing	Librosa, SciPy
NLP & Text Analysis	NLTK, SpaCy, TF-IDF, Word Embeddings
Data Handling	NumPy, Pandas, Matplotlib, Seaborn
Deployment	Streamlit

⚙️ Implementation Workflow

🔹 Deepfake Image Detection

📌 Tech Stack: TensorFlow, OpenCV, Scikit-learn
✔ Collect & preprocess images (FaceForensics++, Celeb-DF, DFDC).
✔ Train CNN models (Xception, EfficientNet, ResNet).
✔ Evaluate accuracy with precision, recall, and F1-score.
✔ Deploy model using Streamlit.

🔹 Deepfake Audio Detection

📌 Tech Stack: Librosa, TensorFlow, Scikit-learn
✔ Collect & preprocess datasets (ASVspoof, FakeAVCeleb).
✔ Extract MFCCs, spectrograms, and chroma features.
✔ Train CNN-based classification models.
✔ Deploy using an interactive web app.

🔹 Fake News Detection

📌 Tech Stack: NLTK, SpaCy, Scikit-learn
✔ Collect & clean text datasets (FakeNewsNet, LIAR, Kaggle Fake News).
✔ Convert text into TF-IDF vectors or embeddings.
✔ Train Logistic Regression classifier.
✔ Evaluate with confusion matrix and performance metrics.
✔ Deploy on Streamlit for real-time analysis.

📊 Performance Metrics

✅ Image Detection Accuracy: 90-98%
✅ Audio Detection Accuracy: 85-95%
✅ Fake News Classification Accuracy: 80-95%
✅ Inference Speed: <1 second per input

🚀 Quick Start Guide

1️⃣ Clone the Repo:

git clone https://github.com/RohanExploit/ShivaSpear_Deepfake.git
cd ShivaSpear_Deepfake

2️⃣ Install Dependencies:

pip install -r requirements.txt

3️⃣ Run the Web App:

streamlit run app.py

4️⃣ Upload an Image, Audio File, or News Text to analyze deepfakes.

📚 References

🔗 FaceForensics++: GitHub
🔗 Celeb-DF Dataset: GitHub
🔗 ASVspoof Dataset: Edinburgh DataShare
🔗 FakeNewsNet: GitHub

👥 Contributors

• Rohan Gaikwad – Project Lead
• Team Members – Devshri Damle , Satyam Kadam 📧 Contact: [itzrohan007@gmail.com] | 🔗 [www.linkedin.com/in/rohanvijaygaikwad]

⭐ If this project helps you, consider starring the repo! ⭐