LifeCount: A Device-free CSI-based Human Counting Solution for Emergency Building Evacuations

Abstract

During large scale building evacuations, it is difficult to ascertain how many people have left the premises safely. To assist in the rescue effort, indoor counting solutions can provide emergency personnel with the number of people who have evacuated the building, and from which floors. LifeCount implements a novel two stage neural network-based algorithm to accurately count the number of people passing through a hallway. Experimental results show that LifeCount can attain a zero counting error accuracy of 96.9%.