Kayak and Sailboat Detection Based on the Improved YOLO with Transformer

In this paper, we aim at sailboat and kayak detection from digital images using deep learning. The main dataset is created by ourselves, we have collected the images at the harbour near our city, Auckland. In the sailboat and kayak detection, we search for a set of best parameters for the baseline of YOLOv5 model. In this paper, we propose a spate of backbone structures for the purpose of comparisons, we are able to find out the best structure for the kayak detection using our training dataset. Finally, we verify our proposed model and compare it with an existing well-trained model by using ensemble learning. All results are obtained based on a computer with GTX1060 6G RAM GPU.