Presence Detection with Wi-Fi Using ESP32

This paper addresses the increasing demand for presence detection across various industries while addressing concerns related to invasive technologies such as cameras and microphones. The focus is on identifying individuals or objects within a specified area without compromising privacy. Introducing a cost and power efficient Wi-Fi system leveraging the ESP32 system-on-chip, with built-in Wi-Fi capabilities, is crucial for achieving a balance between accuracy and affordability in Wi-Fi-based presence detection. The designed system uses CNN (Convolutional Neural Network) for classifying occupancy and demonstrates a high accuracy of around 86% in classifying occupancy counts up to 3 people. Notably, accuracy reaches 96 % for 0–2 people, and presence/absence determination reaches 98 %. Challenges in classifying 0–4 people due to confusion in 0 and 4 people, attributed to the limitations of ESP32 chip and potential over-fitting, provide insights for future improvements. In conclusion, this Wi-Fi-based presence detection system offers a non-intrusive alternative to traditional methods. Positive results underscore its potential to enhance non-invasive monitoring technologies, paving the way for accurate and privacy-respecting solutions in diverse real-world applications.