Forensic Verification to Differentiate Between Real and Fake Images

A Python toolkit for digital image forgery detection focused on three common scenarios:

• Copy–Move (regions copied within the same image) — DBSCAN-based clustering + CFA (Color Filter Array) artifacts
• Splicing (regions pasted from another image) — ELA (Error Level Analysis)
• Double Compression — JPEG double-encoding detection

The system outputs visual heatmaps/masks, quantitative scores, and a short report per image.

---

🔍 Why this matters

Easy-to-use editing tools make tampering common. Verifying image authenticity is vital for law enforcement, multimedia platforms, and digital forensics. This project offers reproducible baselines you can run locally to flag copy–move, splicing, and double-JPEG signatures.

---

✅ Pre-requisites

1. Project folder ready (e.g., Forgery_Detector/) with application files
2. Python installed (3.9–3.12 recommended)
3. Python on PATH
   Windows: This PC → Right Click → Properties → Advanced system settings → Environment Variables → Path → ensure Python is listed
4. Install dependencies with pip from requirements.txt

---

📦 Installation

1) Clone or copy this repo into Forgery_Detector/

cd Forgery_Detector

2) Download CASIA2 dataset and save it in same folder

3) (Recommended) Create & activate a virtual environment

python -m venv .venv

Windows:

.venv\Scripts\activate

macOS/Linux:

source .venv/bin/activate

4) Install dependencies

pip install -r requirements.txt