Overview / Usage

Motivation:

Social distancing is playing an important role during the ongoing COVID-19 pandemic. The first case of a COVID-19 was reported in Wuhan, China. Within 1-month of the outbreak, the number of positive cases and the deceased rose at an exponential rate. Later on, with the implementation of “Social distancing”, the deceased count reduced drastically, and the it was later adopted and implemented worldwide.

What are problems being solved?

Social distancing is a method, primarily used to control the spread of contagious diseases like Covid-19, etc. As the name itself suggests, social distancing indicates that people should physically distance themselves from other people, in order to reduce the close contact, and thereby to reduce the spread of a contagious disease such as Covid-19. Social distancing is disputably the most effective non-pharmaceutical way to prevent the spread of contagious disease such as Covid-19. Social distancing, by definition implies that : if people are not close together, then they cannot spread germs.

Using technology, more specifically the ML-based, I have developed a mechanism, which can be implemented on a video stream of any of the localities. The mechanism we have planned to develop, distinguishes the people who are following social distancing, and those who are not following.

Methodology / Approach

Methodology:

• Detect humans in the frame with yolov3.
• Calculates the distance between every human who is detected in the frame.
• Shows how many people are violating social distancing.

• Python3.x
• Cython==0.29.19
• imutils==0.5.3
• numpy==1.18.4
• opencv-python==4.3.0.36
• Pillow==8.2.0
• PyYAML==5.4
• six==1.15.0
• torch==1.4.0
• torchvision==0.5.0

Pre-trained YOLO3 model for Human detection: “https://pjreddie.com/media/files/yolov3.weights”

Technologies Used

Python, Deep Learning, Computer Vision, OpenCV, Yolo Object Detection

Documents and Presentations

Repository

https://github.com/VinayEdula/social_distancing_system