Marine ship detection method and system based on multi-scale convolutional neural network model

Abstract

The invention provides a marine ship detection method and system based on a multi-scale convolutional neural network model, and the method comprises the steps: building a ship image sample library, collecting the ship video data of a coastal region under visible light based on an unmanned plane platform, extracting each frame of image, and obtaining the true value, length and width of the positionof a ship; performing enhancement processing on the data through digital image processing algorithms such as inversion and zooming; constructing a multi-layer convolutional neural network as a ship target detector, and inputting the obtained processed image as sample data into a deep learning network to obtain a convolutional feature map; and constructing a multi-scale convolutional neural unit,performing feature fusion on the multi-layer convolutional feature map based on the convolutional feature map, and performing training according to the obtained real position of the ship to obtain a training model. Due to the fact that a multi-scale fusion method is adopted, the detection accuracy is well guaranteed, and the training difficulty is reduced.

Description

technical field [0001] The invention belongs to the technical field of ship digital image processing, in particular to a method and system for detecting ships at sea based on a multi-scale convolutional neural network model. Background technique [0002] Surveillance cameras are ubiquitous in today's society. If we only rely on human eye observation and detection, it is easy to miss abnormal events in the video. With the rapid development of computer network, communication and semiconductor technology, people are more and more interested in using computer vision to replace human eyes to analyze video images and obtain useful information. Target detection is a focus of computer vision research, and its main function is to extract the position of the target of interest and other information in the image. Object detection is the basis of many video applications, and it is also a necessity for applications such as traffic monitoring, intelligent robots, and human-computer inter...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that Hough only has a good segmentation effect on straight coastlines, and the segmentation robustness is poor; the Faster RCNN network needs to artificially specify the default frame size of the algorithm in advance, which is difficult to train and relatively difficult. Difficult to accommodate vessels of various sizes

Images