IIC Journal of Innovation 15th Edition | Page 35

Safely Back to Work
Fig . 10 : Positioning the camera ( from surveillancereviews . net ) 16
One approach is that the software running on the computer can use two Deep Learning models 19 ( Multi-layer Neural Network ) that provide intelligence to the system . People in the frame are detected and localized using an object detection 20 model , while a depth estimation 21 model tries to predict the distance of detected people with respect to the position of the camera . This data is passed on to a custom algorithm to conclude if people in the given video frame are following the social-distancing guidance .
Fig . 11 : Process flow for Social Distance Monitoring via Video Surveillance
19 https :// en . wikipedia . org / wiki / Deep _ learning
20 https :// machinelearningmastery . com / object-recognition-with-deep-learning /
21 https :// towardsdatascience . com / depth-estimation-1-basics-and-intuition-86f2c9538 cd1
- 30 - November 2020