Concealed Weapon Detection Using Thermal Cameras.

Area of Science:

• Computer Science
• Electrical Engineering
• Security Technology

Background:

• Increasing security threats necessitate advanced weapon detection.
• Existing methods struggle with concealed weapons and false alarms.

Purpose of the Study:

• To develop a novel two-stage method for concealed handgun detection using thermal imaging and deep learning.
• To create a practical, lightweight system for law enforcement and surveillance applications.

Main Methods:

• A two-stage approach: frame-level firearm detection followed by person association.
• Development of a lightweight algorithm for embedded systems and a tailored thermal dataset.
• Implementation on a chest-worn Android smartphone with a miniature thermal camera for hands-free operation.

Main Results:

• Achieved a mean Average Precision (mAP@50-95) of 64.52% on a custom thermal dataset.
• Significantly reduced false positives and false negatives compared to state-of-the-art methods.
• Demonstrated effectiveness in controlled concealment scenarios.

Conclusions:

• The proposed method offers a scalable and practical solution for concealed weapon detection.
• The system's optimization for low-end embedded devices enables wearable and mobile security applications.
• Enhanced reliability and accuracy provide a valuable tool for law enforcement and surveillance.