Ship Target Detection Method Based on Berthing Scenarios

In port operations, berthing procedures play a crucial role, with ship detection serving as the cornerstone for automation in berthing scenarios. Existing ship detection methods suffer from issues such as low accuracy and slow detection speed. This paper proposes a ship detection method based on the SSD algorithm. The approach leverages dilated convolution to enhance features, utilizes the lightweight MobileV2 network to optimize the model, and ultimately employs the K-means++ method to refine the design of default box ratios. This method achieves high accuracy, rapid detection speed, and a lightweight profile, providing robust technical support for efficient ship target recognition in berthing operations.