Accuracy Comparison of Home Security Face Recognition Model in The Several Lighting Condition Using Some Kinect Produced Image
Octgi Ristya Perdana, Aris Tjahyanto, Febriliyan Samopa · 2021
As technology advances, security systems and the development of artificial intelligence systems around the world continue to be developed. Artificial intelligence is very often used to automate parts of a system. In terms of data security and physical security, the use of artificial intelligence to be able to recognize a person's face plays a very important role in increasing the efficiency and management of knowledge. This paper was written with the aim of proposing a comprehensive comparison of accuracy, in recognizing faces using the features found on the Kinect camera (depth images, 3-dimensional images, and infrared images), compared to features found on ordinary cameras (RGB images), using techniques digital image processing in several different light conditions (dark room conditions and bright / standard room lighting). This research was conducted by comparing how reliably the facial recognition system classifies images that are part of the data class, as well as fake class images in the home security system. From this research, the Support Vector Machine (SVM) algorithm is used for classification, resulting in 20% better accuracy for images produced by Kinect cameras than for standard RGB cameras.