A YOLOv5 Baseline for Underwater Object Detection

Abstract

Underwater object detection is harder than that on land. With the rapid development of deep learning, more and more object detection algorithms have been proposed. They have a great performance in object detection on land, but lack ideal performance in underwater object detection. In the line of this observation, we introduce a YOLOv5 baseline for underwater object detection. YOLOv5 is an extremely fast end-to-end algorithm to detect the objects and it has four sizes of models. This makes it easy to choose the right algorithm according to the underwater equipment. Experimental results show that YOLOv5 is a lightweight, fast and accurate object detection algorithm which is suitable for underwater environment. We use these experimental results as a YOLOv5 baseline for underwater object detection. Other researchers who are interested in underwater object detection can conduct research on this baseline.