Spot-Wise Smart Parking: An Edge-Enabled Architecture with YOLOv11 and Digital Twin Integration

TL;DR

This work tackles the inefficiency of drivers searching for parking by delivering spot-level occupancy using an edge-enabled, spot-wise approach. It enhances a campus parking system with a distance-aware vehicle-to-spot matching method and an Adaptive Bounding Box Partitioning mechanism to robustly assign detections to spots, achieving a balanced accuracy of 0.9880 and maintaining a compact 40.5 MB YOLOv11m-TFLite model, with end-to-end latency suitable for minute-scale updates. A new Digital Shadow component provides standardized visualization and interoperability via NGSI-LD/FIWARE, and an application support server on a repurposed TV Box enables scalable, low-cost cloud integration. The system demonstrates practical occupancy analytics, dashboards, and data-driven insights while outlining a clear path toward a full Smart Campus Digital Twin through standardized data models, bi-directional simulation, and AI-assisted decision support. Overall, the approach yields accurate, scalable, edge-native spot occupancy with actionable insights for campus mobility and urban planning."

Abstract

Smart parking systems help reduce congestion and minimize users' search time, thereby contributing to smart city adoption and enhancing urban mobility. In previous works, we presented a system developed on a university campus to monitor parking availability by estimating the number of free spaces from vehicle counts within a region of interest. Although this approach achieved good accuracy, it restricted the system's ability to provide spot-level insights and support more advanced applications. To overcome this limitation, we extend the system with a spot-wise monitoring strategy based on a distance-aware matching method with spatial tolerance, enhanced through an Adaptive Bounding Box Partitioning method for challenging spaces. The proposed approach achieves a balanced accuracy of 98.80% while maintaining an inference time of 8 seconds on a resource-constrained edge device, enhancing the capabilities of YOLOv11m, a model that has a size of 40.5 MB. In addition, two new components were introduced: (i) a Digital Shadow that visually represents parking lot entities as a base to evolve to a full Digital Twin, and (ii) an application support server based on a repurposed TV box. The latter not only enables scalable communication among cloud services, the parking totem, and a bot that provides detailed spot occupancy statistics, but also promotes hardware reuse as a step towards greater sustainability.

Figures (20)

• Figure 1: IoT layers of the proposed system. Component image sources: https://commons.wikimedia.org/wiki/File:Raspberry_Pi_3_B%2B_%2839906369025%29.png and https://gopigo.io/raspberry-pi-camera/.
• Figure 2: Edge Computing Device Flow. The circle indicates the start of the script and the dashed arrow indicates the loop.
• Figure 3: Labeling of the center coordinates of each spot.
• Figure 4: ROI reference mask used to select the parking lot area.
• Figure 5: Assigning method. The orange line indicate the distance of the center of the detection to the center of the labeled bounding box. The bigger red bounding boxes indicate detected vehicles, while the smaller bounding box is red for spots occupied and green for empty.