The marine environment is becoming increasingly complex due tothe various marine vehicles,and the diversity of maritime objects poses a challengeto marine environmental governance.Maritime object detection technologyp...The marine environment is becoming increasingly complex due tothe various marine vehicles,and the diversity of maritime objects poses a challengeto marine environmental governance.Maritime object detection technologyplays an important role in this segment.In the field of computer vision,there is no sufficiently comprehensive public dataset for maritime objects inthe contrast to the automotive application domain.The existing maritimedatasets either have no bounding boxes(which are made for object classification)or cover limited varieties of maritime objects.To fulfil the vacancy,this paper proposed the Multi-Category Large-Scale Dataset for MaritimeObject Detection(MCMOD)which is collected by 3 onshore video camerasthat capture data under various environmental conditions such as fog,rain,evening,etc.The whole dataset consists of 16,166 labelled images alongwith 98,590 maritime objects which are classified into 10 classes.Comparedwith the existing maritime datasets,MCMOD contains a relatively balancedquantity of objects of different sizes(in the view).To evaluate MCMOD,this paper applied several state-of-the-art object detection approaches fromcomputer vision research on it and compared their performances.Moreover,a comparison between MCMOD and an existing maritime dataset was conducted.Experimental results indicate that the proposed dataset classifies moretypes of maritime objects and covers more small-scale objects,which canfacilitate the trained detectors to recognize more types of maritime objects anddetect maritime objects over a relatively long distance.The obtained resultsalso showthat the adopted approaches need to be further improved to enhancetheir capabilities in the maritime domain.展开更多
基金supported by the Important Science and Technology Project of Hainan Province under Grant(ZDKJ2020010).
文摘The marine environment is becoming increasingly complex due tothe various marine vehicles,and the diversity of maritime objects poses a challengeto marine environmental governance.Maritime object detection technologyplays an important role in this segment.In the field of computer vision,there is no sufficiently comprehensive public dataset for maritime objects inthe contrast to the automotive application domain.The existing maritimedatasets either have no bounding boxes(which are made for object classification)or cover limited varieties of maritime objects.To fulfil the vacancy,this paper proposed the Multi-Category Large-Scale Dataset for MaritimeObject Detection(MCMOD)which is collected by 3 onshore video camerasthat capture data under various environmental conditions such as fog,rain,evening,etc.The whole dataset consists of 16,166 labelled images alongwith 98,590 maritime objects which are classified into 10 classes.Comparedwith the existing maritime datasets,MCMOD contains a relatively balancedquantity of objects of different sizes(in the view).To evaluate MCMOD,this paper applied several state-of-the-art object detection approaches fromcomputer vision research on it and compared their performances.Moreover,a comparison between MCMOD and an existing maritime dataset was conducted.Experimental results indicate that the proposed dataset classifies moretypes of maritime objects and covers more small-scale objects,which canfacilitate the trained detectors to recognize more types of maritime objects anddetect maritime objects over a relatively long distance.The obtained resultsalso showthat the adopted approaches need to be further improved to enhancetheir capabilities in the maritime domain.
基金supported by the Major Research Project of National Natural Science Foundation of China(92061123)the Key Research Program of the Chinese Academy of Sciences(QYZDJ-SSW-SLH01)the Youth Innovation Promotion Association of CAS(2022036)。
