Deep Learning and Geo-Fencing Based Free Parking Spot Detection Using a Camera Mounted on a Truck

The global shipping industry is experiencing rapid growth due to increased demand for commodities driven by e-commerce and digital technology. This growth has created new challenges for the logistics industry, which must improve efficiency, accuracy, and speed. Automation of logistics processes is an important research topic that can help address these challenges by reducing human error, increasing efficiency, and saving labor and energy costs. An important area of focus for logistics yard automation is vehicle management, and understanding parking occupancy is critical to better inventory management and faster operations. This study analyzes existing parking space detection methods and evaluates their feasibility in specific logistics scenarios, identifying new challenges and potential benefits of logistics yard automation. Our proposed solution is based on deep learning and Geofence, which exploits the standardization of parking spaces in the logistics park area and combines a deep learning model to estimate the distance and declination of the exchanged bodies within the screen without the need for additional 3D sensors. Our approach offers significant advantages in terms of deployment cost, operational flexibility, and utilization of additional information. It shows the powerful performance of deep learning in object detection on the one hand, and the great application potential of geo-fencing on the other hand, providing a new way of thinking for future parking spot detection.