Urban road pedestrian detection system integrating IoT technology and multi-sensor data fusion algorithm

With the increase of urban roads, sensors are used to detect the presence and behavior of pedestrians in real-time at intersections, zebra crossings, and other places. However, the detection results of existing pedestrian detection systems have low accuracy and are easily interfered with. To address these issues, research was conducted to optimize the YOLO series algorithms by introducing attention mechanism-Neck module, and sensor network model, ultimately constructing a pedestrian detection system in the Internet of Things. The comparative experiments were conducted in scenarios with pedestrian densities of 0.4, 0.8, 1.2, 1.6, and 2.0 people/m 2 . The accuracy of the pedestrian detection system proposed in the study was 0.965, 0.956, 0.957, 0.954, and 0.948, respectively. Comparative experiments were conducted at temperatures of -90 °C, -60 °C, -30 °C, 0 °C, 30 °C, and 60 °C, and the max root mean square error of the pedestrian detection system introduced in the study was 3.6. In addition, when the humidity was 15%, 30%, 45%, 60%, and 75% respectively, the experiment showed that the pedestrian detection system introduced by the research had the highest accuracy of 0.96 and the max detection distance of 11.0. The proposed system adapts to varying pedestrian density scenarios with relatively stable detection accuracy. Under extreme temperatures, the system maintains controllable error margins and reliable detection performance. Moreover, its accuracy and detection range under humidity variations consistently outperform the comparison system. Consequently, the proposed pedestrian detection system demonstrates strong generalization capabilities, high stability, and anti-interference ability, thereby possessing practical value.