Enteromorpha prolifera detection in complex scenarios based on YOLOv8

As a key player in the regulation of the global ecosystem, the oceans and seas, with their vast space and abundant resources, are of vital importance. However, in recent years, with the development of economy, the frequency of marine pollution has risen, especially the large-scale outbreak of seagrass has brought serious impacts on marine ecology and economic activities. The monitoring and prevention of seagrass disaster has become a common concern for scholars and managers. Traditional ground survey methods are inefficient, so the use of advanced target detection technology has become one of the solutions. In recent years, YOLO, as an advanced target detection algorithm, has performed well in various image recognition tasks, but specific comparative studies in seashore moss detection are still relatively few. In this study, we address the problem of seashore detection by comparing their performances in YOLOv5 and YOLOv8 accuracy and detection speed through label transformation and model training on the FIO-EP seashore dataset. The experimental results show that YOLOv8’s recognition accuracy is 4% higher than YOLOv5, and its detection speed is also 0.03 seconds faster. This indicates that YOLOv8 outperforms YOLOv5 in both accuracy and detection speed. This study not only validates the application potential of the YOLO series in monitoring seaweed but also provides valuable references for future related research. This study not only verifies the potential application of YOLO series in seashore monitoring, but also provides a valuable reference for future related research.