Author(s)

Zhiqiang Zhao

Corresponding Author

Zhiqiang Zhao

Abstract

Aiming at the problem of low accuracy of traditional image processing methods in the detection of sulfur liquid in the on-site environment with complex equipment structure, many types of debris and serious ground wear in ships, this paper proposes a sulfur liquid detection algorithm based on CNN. By analyzing the problem of sulfur liquid detection, making a data set, establishing the VGG16 model and combining the training samples of the early stop algorithm to avoid the over-fitting state, the rapid automatic detection of sulfur liquid in complex pipelines is realized. This method can be accurate and accurate in the ship site. Identify sulfur liquid and reduce the influence of noise interference. Finally, the superiority of the algorithm in this paper is verified by comparing with various image processing methods. The results show that the test accuracy of the algorithm can reach 99.44%, and the prediction accuracy can reach 97.0%. Compared with the accuracy of traditional image processing algorithms, and the prediction time of a single image is about 0.2 s, it can meet the detection needs of the ship site.

Keywords

convolutional neural network; ships; sulfur emissions; visual monitoring