My Academic minor project

Using MediaPipe Pose, YOLO v6, and StowFast to Detect Threats in Malls

This project was developed as part of my academic minor project. The goal was to create a real-time threat detection system for shopping malls using computer vision technologies.

The Challenge

Shopping malls face various security challenges, from theft to more serious threats. Traditional security measures often rely on human operators monitoring CCTV feeds, which can be inefficient and prone to human error.

The Solution

We developed a system that combines:

• MediaPipe Pose: For real-time human pose estimation
• YOLO v6: For object detection and classification
• StowFast: For efficient data processing and storage

Implementation Details

The system processes video feeds in real-time, identifying potential threats based on predefined patterns of behavior. When a threat is detected, the system alerts security personnel and provides relevant information about the situation.

Results

The prototype demonstrated significant improvements in threat detection accuracy and response time compared to traditional methods. The system was able to identify various threat scenarios with a high degree of accuracy.

Future Improvements

We plan to enhance the system with:

• Improved accuracy through additional training data
• Integration with existing security systems
• Mobile application for security personnel