Security Guards Image Dataset

Facial recognition plays a critical role in today’s security and surveillance systems, by enabling real-time identification, access control, and anomaly detection in constantly changing environments. But for these systems to perform effectively, there is a need for the availability of diverse and application-specific datasets.
This dataset, titled the Security Guards Facial Image Dataset, presents a collection of 21,034 images distributed as five distinct security guards:
Security guard 1 - 3,934 images
Security guard 2 - 3,794 images
Security guard 3 - 2,475 images
Security guard 4 - 5,731 images
Security guard 5 - 5,100 images

Devices Used:
Oppo A55 (Android) - 50 MP, f/1.8, 1/2.76" sensor size
iPhone 14 Pro Max (iOS) - 48 MP, f/1.78, 24mm (wide), 1/1.28" sensor size

Device Distribution:
iOS: 56.76% images
Android: 43.24% images

Recording Conditions:
Indoor (flash) and outdoor (natural light) environments
4 - 5 unique backgrounds
With and without headgear (cap)
No constraints on facial hair, expression

The dataset offers an organized collection of facial images of security guards, captured in real-world campus settings, to support meaningful research in intrusion detection, and AI-based monitoring. It presents a robust and practical resource for training and validating computer vision models in security applications.

This dataset supports both individual identity recognition and multi-class facial classification tasks. It is well-suited for training and evaluating deep learning models such as convolutional neural networks (CNNs). It allows researchers to enhance the performance and reliability of AI systems deployed in real-time surveillance and security scenarios.