Ship detection based on deep learning under complex lighting

In order to improve the real-time performance and detection rate of ship detection in a complicated marine lighting environment. An improved YOLOv3 algorithm is proposed. This paper first constructs a ship target data set under different light intensities. Then, the problem of poor detection of ship targets caused by the influence of illumination is addressed. The channel attention mechanism module is introduced into the network of the YOLOv3 algorithm. Generate weighting coefficients in the feature channel and enhance the feature extraction capabilities of the network. Finally, the accuracy of detection under complex lighting conditions is improved. The detection results show that the detection accuracy of the YOLOv3 network with the attention mechanism module has been significantly improved in the scene of changing lighting. It has laid a reliable foundation for the target tracking of ships at sea.