SOCIAL DISTANCING DETECTION FINDING OPTIMAL ANGLE WITH YOLO V3 DEEP LEARNING METHOD

Mutiara Dwi Anggraini (Master of Information Engineering, Department Amikom University, Yogyakarta, Indonesia)
Kusrini Kusrini (Master of Information Engineering, Department Amikom University, Yogyakarta, Indonesia)
Hanif Al Fatta (Master of Information Engineering, Department Amikom University, Yogyakarta, Indonesia)

Jurnal Teknik Informatika (JUTIF)
Vol. 3 No. 5 (2022): JUTIF Volume 3, Number 5, October 2022

The COVID-19 pandemic has had a profound impact on all aspects of society. One of the implementations that the government has so far carried out is using masks and maintaining social distance. Over time, social distancing is difficult to control because people are now getting booster vaccines, but some have not. One way to overcome this problem is with a social distance detector system that detects the number of people and the distance of human objects from one another in an area. This study aims to apply in the office area, or the public. This research is one of the developments of a social distancing detector application that produces an optimal angle in using the application. So the program can detect the entirety of the object with optimal accuracy. Angle is very influential in taking the image to be processed in the system. This study uses the python language with the YOLOv3 library. This study got the best results,and the mean average precision in 90%:10% didapatkan dengan learning rate 0,001 dengan nilai mAP 54,11%, deteksi pada saaat penggujian sebesar 100%. Percobaan sudut terdapat 00.150.300, 450.600 dengan total 50 data video testing. Sudut optimal yang didapatkan pada penelitian ini adalah 00.150.300. Hal ini membuktikan bahwa sudut pengambilan video atau peletakan kamera sistem social distancing.