An Analysis Of Maritime Distressing Aircrafts Detection Method Based On YOLO

Thanks to the weather conditions and traditional searching methods,distressing aircrafts detection method has become a critical factor in maritime search and rescue.This article studies the drone application and deep learning in maritime search,and establishes and analyzes a distressing aircrafts detecting method which is based on YOLO using convolutional neural network.As traditional maritime distressing aircrafts detection methods are still using visual search and electronic search,which lacks high precision and speed,this paper proposes a higher efficient and more diversified detection method based on deep learning.Through pattern recognition,this method uses YOLOV4 as the first detecting algorithm and YOLOV4-Tiny as the second algorithm to promote precision and speed in maritime search.Due to the target’s focus feature in maritime search,three-dimensional object takes the place of one-dimensional object,which utilizes aircraft,helicopter,and ship,all with different dimensions,as targets when constructing the detection method so that searching efficiency will be improved.Raw dataset will be constructed by using original image’s physical layer to compensate the traditional detection method’s drawback,inadequate information.With plenty of diverse aspect’s original remote sensing images from narrow angle in the dataset,various combinations will be set up based on object’s feature to ensure convolutional neural network’s learning ability.Analysis of the results of this detection methods shows its precision and speed is 22.01% and 320% better than that of YOLOV3 respectively,confirming this method’s feasibility and effectiveness.Ways and means on dispatching search and rescue resources based on its outcome will be discussed.