AI Image Forensics Web Application Project

AI Image Forensics Web Application

Project Overview A professional-grade, multimodal web application designed to detect and analyze AI-generated visual content. By combining traditional image processing techniques with advanced Large Vision-Language Models (VLMs), the platform provides comprehensive, automated analyses of image authenticity, digital signatures, and physical integrity.

Core Technologies & Frameworks

Next.js (React)

Python, FastAPI

Google Gemini Vision API

Docker, Google Cloud Run

Key Features & Engineering Highlights

8-Stage Forensic Pipeline: Engineered a highly robust image analysis pipeline that sequentially executes metadata scanning, quadrant tiling, Error Level Analysis (ELA), and Canny Edge detection.

Multimodal AI Analysis: Integrated Google Gemini Vision as an automated "forensic analyst" to evaluate visual inputs for physical inconsistencies, blending errors, and AI generation artifacts.

Decoupled Microservices Architecture: Built a high-performance Next.js frontend paired with a highly scalable Python/FastAPI backend to efficiently process complex mathematical image transformations.

Cloud-Native Deployment: Established automated workflows for containerizing the application with Docker and deploying the backend services to Google Cloud Run for scalable compute.

Data Visualization UI: Developed an intuitive, user-friendly dashboard to clearly present complex forensic findings, edge maps, and AI confidence scores to end-users.