ANALYSIS IN DISTANCE MEASUREMENT AND 45 DEGREES IMPLEMENTATION ON PHYSICAL DISTANCING ALERTING SYSTEM BASED OF VIDEO ANALYTICS USING YOLOV3

• Main Author: Naufal Kurniawan, M.
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/55937
• Institution: Institut Teknologi Bandung

Covid-19 are still being one of the main problems in Indonesia. The pandemic which started in March 2020 has changed the people’s daily life drastically until now. As of now, there are over five thousand new Covid-19 cases each day and there’re no signs of stopping. This could be caused by the people’s ignorance to it. There are multiple cases found where people don’t wear mask and/or don’t keep their distance of each other. In this project, the writer proposed a physical distancing alerting system as a mean to bring justice on the ongoing physical distancing policy. This system is built around NVIDIA Jetson Nano, a microcontroller that’s flexible in size and able to run machine learning and video processing processes. The main program is created in Python language and make use of OpenCV, streamlit, Pydub, and gstreamer libraries. In this report, the writer will dive into some of its subsystem which consist of distance measurement system and alarm system. This report will also experiment the system placed in the corner of a closed room, slating for 45 degrees. The distance measurement system is implemented by using two dimensional Euclidean distance algorithm and the alarm system is implemented by Pydub. Experiments shown that the distance measurement system is best suited when the people detected are shown clearly and in a one-lined formation. Other experiments result in an error due to its two dimensional algorithm. The alarm system is working with clear sounds but with a 25 seconds delay. Overall, the system will run more effective in a 45 degrees position than in a 0 degrees position from a human perspective.