A Neural Network based Social Distance Detection

The global breakout of the most spreading viruses like COVID-19 has fundamentally changed the way of interacting with one another. By avoiding people's physical contact, the spread of virus can be lowered. So, one such measure called social distancing can minimize the reproduction rate (RO) of viruses among communities. This paper proposes a system that can automatically estimate the interpersonal distance between people using MobileNet SSD deep learning algorithms. The proposed model first detects the persons in a video stream, and then, it characterizes people based on their distance using different color indications. To make people aware about the social distance violation, this model also provides alert signal. The proposed system allows a quick monitoring of individuals potential behavior to avoid health risk, especially in current pandemic situation. The proposed system can also be used in surveillance systems to alert the government in harmful situations by analyzing people's movement.