Using Real-Time Detection Transformer in Park Management

Object detection is one of the primary tasks in computer vision and a crucial tool for digital governance in smart cities.The YOLO (You Only Look Once) series of object detection techniques have been regarded as efficient methods.With the rise of Transformers, research in various tasks has stirred waves in the field of computer vision.RTDETR (Real Time Detection Transformer) emerges as a derivative method in the object detection domain.It is well known that each detection method has its own characteristics, and selecting an appropriate model according to task requirements is a common practice for executing the task.Therefore, this study retrained YOLOv8-x, YOLOv8-l, RTDETR-x, and RTDETR-l models separately for object categories required by our park management.The results show that in terms of detection accuracy, the YOLOv8-x and RTDETR-x models outperform the YOLOv8-l and RTDETR-l models, and the RTDETR-x demonstrates the best object detection capability.