The northern slope region of the South China Sea(SCS) is a biological hot spot characterized by high primary productivity and biomasses transported by cross-shelf currents, which support the spawning and growth of com...The northern slope region of the South China Sea(SCS) is a biological hot spot characterized by high primary productivity and biomasses transported by cross-shelf currents, which support the spawning and growth of commercially and ecologically important fish species. To understand the physical and biogeochemical processes that promote the high primary production of this region, we conducted a cruise from June 10 and July 2, 2015. In this study, we used fuzzy cluster analysis and optimum multiparameter analysis methods to analyze the hydrographic data collected during the cruise to determine the compositions of the upper 55-m water masses on the SCS northern slope and thereby elucidate the cross-slope transport of shelf water(SHW) and the intrusions of Kuroshio water(KW). We also analyzed the geostrophic currents derived from acoustic Doppler current profiler measurements and satellite data. The results reveal the surface waters on the northern slope of the SCS to be primarily composed of waters originating from South China Sea water(SCSW), KW, and SHW. The SCSW dominated a majority of the study region at percentages ranging between 60% and 100%. We found a strong cross-slope current with speeds greater than 50 cms^(-1) to have carried SHW into and through the surveyed slope area, and KW to have intruded onto the slope via mesoscale eddies, thereby dominating the southwestern section of the study area.展开更多
Image based individual dairy cattle recognition has gained much attention recently. In order to further improve the accuracy of individual dairy cattle recognition, an algorithm based on deep convolutional neural netw...Image based individual dairy cattle recognition has gained much attention recently. In order to further improve the accuracy of individual dairy cattle recognition, an algorithm based on deep convolutional neural network( DCNN) is proposed in this paper,which enables automatic feature extraction and classification that outperforms traditional hand craft features. Through making multigroup comparison experiments including different network layers,different sizes of convolution kernel and different feature dimensions in full connection layer,we demonstrate that the proposed method is suitable for dairy cattle classification. The experimental results show that the accuracy is significantly higher compared to two traditional image processing algorithms: scale invariant feature transform( SIFT) algorithm and bag of feature( BOF) model.展开更多
基金supported by the National Basic Research Program of China (No. 2014CB441500)the National Natural Science Foundation of China (No. 41406021)
文摘The northern slope region of the South China Sea(SCS) is a biological hot spot characterized by high primary productivity and biomasses transported by cross-shelf currents, which support the spawning and growth of commercially and ecologically important fish species. To understand the physical and biogeochemical processes that promote the high primary production of this region, we conducted a cruise from June 10 and July 2, 2015. In this study, we used fuzzy cluster analysis and optimum multiparameter analysis methods to analyze the hydrographic data collected during the cruise to determine the compositions of the upper 55-m water masses on the SCS northern slope and thereby elucidate the cross-slope transport of shelf water(SHW) and the intrusions of Kuroshio water(KW). We also analyzed the geostrophic currents derived from acoustic Doppler current profiler measurements and satellite data. The results reveal the surface waters on the northern slope of the SCS to be primarily composed of waters originating from South China Sea water(SCSW), KW, and SHW. The SCSW dominated a majority of the study region at percentages ranging between 60% and 100%. We found a strong cross-slope current with speeds greater than 50 cms^(-1) to have carried SHW into and through the surveyed slope area, and KW to have intruded onto the slope via mesoscale eddies, thereby dominating the southwestern section of the study area.
基金Science and Technology Support Plan Project of Tianjin Municipal Science and Technology Commission(No.15ZCZDNC00130)
文摘Image based individual dairy cattle recognition has gained much attention recently. In order to further improve the accuracy of individual dairy cattle recognition, an algorithm based on deep convolutional neural network( DCNN) is proposed in this paper,which enables automatic feature extraction and classification that outperforms traditional hand craft features. Through making multigroup comparison experiments including different network layers,different sizes of convolution kernel and different feature dimensions in full connection layer,we demonstrate that the proposed method is suitable for dairy cattle classification. The experimental results show that the accuracy is significantly higher compared to two traditional image processing algorithms: scale invariant feature transform( SIFT) algorithm and bag of feature( BOF) model.
