Skip to main content
NUS Project

Violence Detection

Imagine a security system so smart, it spots trouble before your nosy neighbor does. With real-time alerts, location tracking, and even a friendly Telegram message to authorities, this project is the ultimate tattletale for crime. Plus, it works just as well at night—because criminals don’t respect bedtime!

To see how its done click here!

Alertify is a cutting-edge surveillance system designed to detect and report violent activities in real time, ensuring faster responses and enhanced public safety. It combines advanced machine learning, computer vision, and cloud analytics to create a reliable and efficient solution for monitoring high-risk areas.

Key Features:

  1. Violence Detection:

    • Achieved 95% accuracy in detecting violent incidents, using a fine-tuned MobileNetV2 model with a custom self-learning rate algorithm.

    • Automatically alerts authorities via Telegram, providing critical details such as crime location, time, and images for immediate action.

  2. Enhanced Nighttime Detection:

    • Integrated Deep Sort tracking to improve object tracking and optimize nighttime detection using computer vision techniques.

    • Achieved a 30% boost in low-light accuracy, ensuring reliable performance regardless of lighting conditions.

  3. Cloud Expertise and Data Processing:

    • Leveraged AWS Data Analytics and Big Data tools during an internship to streamline real-time data processing.

    • Enhanced system responsiveness by efficiently handling large-scale surveillance data.

  4. Efficiency Improvements:

    • Increased model efficiency by 25% through optimized algorithmic design, ensuring quick and accurate incident detection.

Alertify doesn’t just keep an eye out—it ensures authorities are always in the loop, whether it’s day or night. With its advanced features and smart integrations, this system is the silent guardian every community needs.