Offshore wind power safety monitoring and early warning method based on machine vision

Abstract

The invention discloses an offshore wind power safety monitoring and early warning method based on machine vision, and the method comprises the following steps: carrying out the preprocessing of a collected ship image, and building a ship detection data set containing a plurality of types; s2, a YOLOv5 network is used as a reference network, a backbone network, a Neck structure and a detection head of the YOLOv5 network are improved, an improved YOLOv5 ship detection network is constructed, the ship detection network is trained through the data set obtained in the step S1, and a ship detection inference model is obtained; and inputting a ship picture or video stream which needs to be detected and identified into the ship detection reasoning model to obtain the category of the ship and frame out the coordinate position, and monitoring and early warning the ship which affects the offshore wind power safety. The method is suitable for solving a small target detection task under complex background interference, and can improve the sensitivity of the model to targets of different sizes and reduce the omission ratio of smaller targets on the premise of ensuring real-time detection.

Description

technical field [0001] The invention belongs to the field of deep learning and target detection, and in particular relates to a computer vision-based method for identifying and locating objects at sea. Background technique [0002] Wind power is a renewable energy with strong competitiveness and fast speed. Compared with land-based wind energy, the advantages of offshore wind energy resources are mainly stable wind direction, high wind speed, and less impact on the environment. At the same time, offshore wind farms are closer to cities with high energy demand, and the sea surface can be applied to a vast area, which has become the main use of wind power development. Trend, it is predicted that the global offshore wind power scale will increase by 15 times in 2040, when China's wind power installed capacity is expected to increase to 110GW. At present, the development and construction of offshore wind power in my country is in full swing, and a large number of wind farm proj...

AI Technical Summary

Problems solved by technology

Multi-stage detectors have high target detection accuracy, while single-stage detectors have faster detection speed and higher scalability, but there are also shortcomings, that is, the detection effect on small targets is always not ideal

[0007] With the update of the YOLO series, the YOLOv5 algorithm currently has excellent detection accuracy and speed in the field of target detection. However, the maritime ship data itself has the characteristics of dense multi-scale, easy to be occluded, and complex detection scenes, which lead to this method in real-time detection. Under these conditions, the performance of ship small target extraction is not very satisfactory, and problems such as low accuracy of recognition results, missed detection and false detection often occur, and it is difficult to form a complete ship target recognition and positioning method

Images