Secure Video loT Monitoring System Based on ESP32

Monitoring systems have become an integral part of modern households and offices. Recent advancements in loT technologies have made highly sophisticated systems affordable. The use of Artificial Intelligence in monitoring systems, especially with cameras, has become more accessible and sought-after. Currently, there are systems available for performing people-counting tasks in buildings, but they often come with a high cost and high power usage. Images of people as well as the count of people in a room is sensitive data in many occasions. One of potential solutions could be usage of esp32 based cameras due to low cost, small power consumption and compact design. Important to note that esp32 has limited computing resources and is not able to process images locally with AI tools. At the same time, secure data transfer also requires resources to encrypt data due to that available resolution from camera could be reduced. In this paper, the authors investigate the limitations of ESP32based cameras for people counting through a series of tests involving different ESP32-based cameras and configurations. To demonstrate the proof-of-concept, a central computing module was employed in conjunction with MQTT-based TLS secured communication.The results indicate that this approach could achieve a maximum resolution of up to 1600x1200, with a potential image transmission frequency of up to 20 images per minute in some cases.