Combining camera traps and artificial intelligence for monitoring visitor frequencies in natural areas: Lessons from a case study in the Belgian Ardenne

英语. peer reviewed

英语. Visitor monitoring is essential for ecosystem management and the evaluation of ecosystem services. However, in natural areas without entrance fees and with scattered entry and exit points, this task can be challenging, costly, and labor-intensive. Camera traps can provide both quantitative and qualitative data on visitor frequencies, profiles, and activities in these remote areas. Manual image analysis, however, is time-consuming when dealing with large datasets. In this study, we analyzed more than 700,000 images collected by nineteen cameras over a year on hiking trails in the Belgian Ardenne. Consistent with recent studies, our research demonstrates that the use of a convolutional neural network (CNN) can achieve accurate and promising results in detecting and classifying people and non-people (dogs, bicycles). Nevertheless, automatic processing entails the risk of multiple counts of the same individuals, depending on camera’s position, technical characteristics, and the time intervals between photos. This paper discusses the limitations and potential improvements of the monitoring methodol ogy, from camera setup to data analysis. It concludes by the added value of this approach for the management of natural areas. Management implications: The integration of AI with camera traps offers a practical and scalable solution for natural areas management by providing accurate data on visitor frequencies and behaviors. This approach can help site managers optimize visitor flows, reduce the impact of human activities on vulnerable ecosystems, and address user conflicts. It also supports sustainable tourism by informing decisions related to infrastructure, conservation priorities, and visitor access. Additionally, the flexibility of this method allows for site-specific adaptations, ensuring that monitoring efforts are aligned with management objectives while maintaining data transparency and privacy protection.