Pedestrian flow monitoring plays a vital role in urban planning, transportation management, and smart city or tourism applications. However, monitoring is often needed in challenging scenarios with spatial constraints, low traffic, no power supply, and poor network connectivity, such as birdwatching observatories or open-air heritage sites. This paper presents a cost-effective, energy-efficient pedestrian counting system tailored for such isolated locations. The system uses two pairs of active infrared (AIR) sensors to count individuals and determine movement direction. To reduce energy consumption, passive infrared (PIR) sensors are used to trigger both the AIR detectors and an IoT board that handles local processing and communication. Depending on local conditions, LoRaWAN or Wi-Fi is used for uplink communication, allowing flexible deployment. The architecture is built with open-source software and off-the-shelf hardware, supporting low-cost real-world implementation. A comparative analysis of several AIR detectors was conducted, assessing power consumption, detection accuracy, susceptibility to interference, and horizontal beam angle. This fills a gap in the literature regarding the systematic evaluation of AIR sensors. Results highlight the effectiveness of the proposed system in achieving both accuracy and energy efficiency, supporting non-intrusive pedestrian monitoring in various smart city and tourism contexts. The system is currently being deployed and evaluated in the Tagus Estuary Bird Observation and Conservation Area (EVOA).

This work was partially funded by the MoniCrowd project (DOI: 10.54499/2024.07624.IACDC), supported by measure “RE-C05-i08.M04 of the Recovery and Resilience Plan (PRR) and by FCT/MECI and EU funds under UID/50008: Instituto de Telecomunicações.

Wireless sensor networks,Pedestrians,Accuracy,Smart cities,Detectors,Sensor systems,Energy efficiency,Monitoring,Intelligent sensors,Wireless fidelity,Infrared Sensor Comparison,Pedestrian Counting,Low-Power Sensor Networks

• Ciências da Computação e da Informação - Ciências Naturais