Integration of Deep Learning YOLOv3 and OpenCV Methods for Physical Distancing in Covid19 Transmission Prevention

Abstract

The current direction of research on the use of sensors and electronic components is a major topic, even the emergence of the concept of machine learning makes the priority of research development in the field of electronics sharper and focuses on algorithms, more complex methods, and sensor capabilities. This research will raise the topic of preventing the spread of the Covid19 virus. The problem in this research is to determine the right algorithm to engineer electronically, so this research chooses the Deep Learning OpenCV and YOLOv3 methods built with Raspberry Pi and Phyton language based on machine learning to answer hypotheses related to preventing the spread of the Covid19 virus starting from implementing an object recognition system, to calculating real-time physical distancing so that the purpose of this research can display object recognition that can measure physical distancing. From data collection and problems, this research designs a prototype physical distancing tool using OpenCV with the Yolov3 Deep Learning Method. The output of this research is image capture in .jpg and .jpeg formats. Capture of image objects in observation using clear perspective images so that the program can recognize image object captures with red marking when violating physical distancing and this prototype will work if the objects are close to each other in response to detection with an average response time of 9.525133 second Physical Distancing. This is influenced by the size of the file and the number of objects captured.