Back to Projects

Privify

Privacy-focused AI tool for detecting and removing sensitive content from images and videos.

Project Overview

Privify is a privacy-focused AI tool that detects and removes sensitive content from images and videos before they are shared online. It reads metadata, performs entity and content analysis using LLMs, and blurs/redacts private data like faces, license plates, IDs, documents, or geo-coordinates to ensure safe sharing.

Key Features

  • Sensitive content detection (faces, plates, IDs, etc.)
  • Metadata and intent analysis with LLMs
  • Blurring/redaction of private data
  • No storage for privacy
  • Video/image processing with ffmpeg and OpenCV
  • Azure and Vercel deployment

Case Study

Problem

Sharing images and videos online can inadvertently expose sensitive information, leading to privacy breaches and misuse of personal data.

Solution

Privify automatically detects and redacts sensitive content from media files before sharing. It uses a combination of computer vision models and LLMs to analyze both visual and metadata content, ensuring privacy by design.

Research & Techniques

  • Object Detection: Used YOLOv8, OpenCV, and Mediapipe for detecting faces, license plates, and other sensitive entities.
  • Metadata Analysis: Leveraged ExifTool and custom parsers to identify and redact geo-coordinates and document info.
  • LLM-based Intent Analysis: Used Llama 3.1 (Groq) for understanding context and intent in media sharing.
  • Privacy by Design: No user data is stored; all processing is ephemeral and secure.

    • Results

  • Prevented accidental sharing of private data in pilot tests.
  • Adopted by privacy-conscious users and organizations.
  • Recognized as a research demo in privacy-preserving AI.

    • Related Projects

    Smart Hire AI

    Smart Hire AI

    AI-powered recruitment platform with active enterprise pilots. Best FYP Award — UMT Software Engineering Department 2026.

    Vibe Terminal

    Vibe Terminal

    Cross-platform AI terminal assistant — execute shell commands from natural language, with voice I/O and multi-step agent workflows.

    Smart Grade AI

    Smart Grade AI

    AI-powered LMS with RAG-based mentoring — Microsoft Imagine Cup 2025 Global Semifinalist, backed by $25K Azure credits from Microsoft for Startups, in institutional rollout at UMT.