Intelligent Lighting Control under AI Architecture
Song-Bor Chiang, Chih-Hsuan Tsuei · 2023
This study utilized an optimized YOLOv3 training model to improve the accuracy of image recognition technology to over 90% through the data results obtained from the grayscale image and personnel recognition technology of the surveillance system at the experimental field - Baoshan Library. In addition, the system's control screen, combined with the verification field's system architecture, achieved the function of detecting and lighting fixtures within 1.4 to 1.6 seconds when personnel appeared within the detection system range, and the system could turn off the lighting fixtures within 1.4 to 1.9 seconds when personnel disappeared from the detection system.