Improving video surveillance systems in banks using deep learning techniques

Area of Science:

• Computer Science
• Artificial Intelligence
• Security Systems

Background:

• Security and safety are critical for national economics, tourism, and investment.
• Manual surveillance by guards is exhaustive and lacks real-time response capabilities for preventing armed robberies.
• Automated systems are needed for effective crime prevention in public and private spaces.

Purpose of the Study:

• To develop and evaluate a real-time weapon auto-detection framework for video surveillance systems.
• To enhance security by enabling prompt responses to potential threats.
• To minimize false alarms for practical deployment in real-world applications.

Main Methods:

• Implementation of state-of-the-art object detection algorithms, specifically YOLO and Single Shot Multi-Box Detector (SSD).
• Focus on real-time processing for immediate threat identification.
• Model refinement to reduce false alarm rates.

Main Results:

• The proposed framework demonstrates effective real-time weapon detection capabilities.
• The system is designed to minimize false positives, increasing its reliability for practical use.
• The model shows suitability for diverse indoor surveillance environments.

Conclusions:

• The developed early weapon detection system offers a viable solution for enhancing security in various settings.
• The framework can be deployed in both indoor and outdoor surveillance to proactively prevent criminal activities like robberies.
• This technology contributes to improved safety and security through automated threat identification.