Aiming at the environments such as ports with dense ship traffic and sea voyages with complex environments, this paper proposes an efficient YOLOv8 ship detection method based on lightweight improvement. Firstly, the CA attention module is added to the original network structure of YOLOv8, which helps the model to perform feature extraction and target localization more accurately; secondly, the DualConv module is introduced to establish a lightweight network, which optimizes the process of information processing and reduces the amount of floating-point computation efficiently; the experimental results show that the lightweight algorithm achieves an accuracy of 98.7%, and at the same time the algorithm has a GFLOPs of 7.5, which is 8.5% lower than the original model, and the Parameters of the model is 2.65M, which is 12% lower than the original model. It can better meet the demand of rapid ship detection in the harbor site and provides an effective solution for real-time monitoring in the maritime field.
Ship Detection Based on Improved YOLOv8 Algorithm
2024-07-05
976349 byte
Conference paper
Electronic Resource
English