Inductive decision based Real Time Occupancy detector in University Buildings

TL;DR

This paper proposes a real-time occupancy detection system for university buildings using a low-cost sensor setup and a decision tree classifier, aiming to automate manual attendance processes.

Contribution

It introduces a novel occupancy detection method based on reverberation time and multiple sensors, improving real-time monitoring in college environments.

Findings

Reverberation time is a key feature for occupancy detection.

Adding more sensor data decreased classification accuracy.

The system enables centralized real-time occupancy monitoring.

Abstract

The ability to estimate College Campus Occupancy for Classrooms and Labs in real time has become one of the major concerns for various Academicians, authorities and administrators,where still a manual attendance marking system is being followed. Using a low budget multiple sensor setup installed in a college auditorium, the goal is to build a real-time occupancy detector. This paper presents an Inductive real time Decision tree based classifier using multiple sensor dataset to detect occupancy. Using simple feature based thresholds, Reverberation time which comes out to be a novel as well as most distinguishing feature sampled at various frequencies over a given time interval was used to detect the occupancy with an accuracy of %.Addition of various other sensor data, decreased the accuracy of classification results. The detector setup can be used in various college buildings to provide real time centralised occupancy status thus automating the manual attendance system being used.