A new highly oxygenated guaiane lactone 5-epi-menverin C(1),along with four known compounds including one ses-quiterpene(2)and three steroids(3-5),were isolated from gorgonian Echinomuricea indomalaccensis which were ...A new highly oxygenated guaiane lactone 5-epi-menverin C(1),along with four known compounds including one ses-quiterpene(2)and three steroids(3-5),were isolated from gorgonian Echinomuricea indomalaccensis which were collected from the South China Sea.Through anlyzing the NMR data and comparing with other reported compounds,their structures were determined.The absolute configuration of compound 1 was determined by comparing its experimental ECD with that obtained by calculation.In a bioassay in vitro,5-epi-menverin C(1)displayed antiviral activity against the HSV-1 virus with an inhibition rate of 69.2%(c=25μmol L−1).展开更多
目的在近岸合成孔径雷达(synthetic aperture radar,SAR)图像舰船检测中,由于陆地建筑及岛屿等复杂背景的影响,小型舰船与周边相似建筑及岛屿容易混淆。现有方法通常使用固定大小的方形卷积核提取图像特征。但是小型舰船在图像中占比较...目的在近岸合成孔径雷达(synthetic aperture radar,SAR)图像舰船检测中,由于陆地建筑及岛屿等复杂背景的影响,小型舰船与周边相似建筑及岛屿容易混淆。现有方法通常使用固定大小的方形卷积核提取图像特征。但是小型舰船在图像中占比较小,且呈长条形倾斜分布。固定大小的方形卷积核引入了过多背景信息,对分类造成干扰。为此,本文针对SAR图像舰船目标提出一种基于可变形空洞卷积的骨干网络。方法首先用可变形空洞卷积核代替传统卷积核,使提取特征位置更贴合目标形状,强化对舰船目标本身区域和边缘特征的提取能力,减少背景信息提取。然后提出3通道混合注意力机制来加强局部细节信息提取,突出小型舰船与暗礁、岛屿等的差异性,提高模型细分类效果。结果在SAR图像舰船数据集HRSID(high-resolution SAR images dataset)上的实验结果表明,本文方法应用在Cascade-RCNN(cascade region convolutional neural network)、YOLOv4(you only look once v4)和BorderDet(border detection)3种检测模型上,与原模型相比,对小型舰船的检测精度分别提高了3.5%、2.6%和2.9%,总体精度达到89.9%。在SSDD(SAR ship detection dataset)数据集上的总体精度达到95.9%,优于现有方法。结论本文通过改进骨干网络,使模型能够改变卷积核形状和大小,集中获取目标信息,抑制背景信息干扰,有效降低了SAR图像近岸复杂背景下小型舰船的误检漏检情况。展开更多
基金supported by the National Key Research and Development Program of China (No. 2018YF C0310900)the National Natural Science Foundation of China (No. 41830535)+1 种基金the Fundamental Research Funds for the Central Universities of China (No. 201962002)the Taishan Scholars Program, China
文摘A new highly oxygenated guaiane lactone 5-epi-menverin C(1),along with four known compounds including one ses-quiterpene(2)and three steroids(3-5),were isolated from gorgonian Echinomuricea indomalaccensis which were collected from the South China Sea.Through anlyzing the NMR data and comparing with other reported compounds,their structures were determined.The absolute configuration of compound 1 was determined by comparing its experimental ECD with that obtained by calculation.In a bioassay in vitro,5-epi-menverin C(1)displayed antiviral activity against the HSV-1 virus with an inhibition rate of 69.2%(c=25μmol L−1).
文摘目的在近岸合成孔径雷达(synthetic aperture radar,SAR)图像舰船检测中,由于陆地建筑及岛屿等复杂背景的影响,小型舰船与周边相似建筑及岛屿容易混淆。现有方法通常使用固定大小的方形卷积核提取图像特征。但是小型舰船在图像中占比较小,且呈长条形倾斜分布。固定大小的方形卷积核引入了过多背景信息,对分类造成干扰。为此,本文针对SAR图像舰船目标提出一种基于可变形空洞卷积的骨干网络。方法首先用可变形空洞卷积核代替传统卷积核,使提取特征位置更贴合目标形状,强化对舰船目标本身区域和边缘特征的提取能力,减少背景信息提取。然后提出3通道混合注意力机制来加强局部细节信息提取,突出小型舰船与暗礁、岛屿等的差异性,提高模型细分类效果。结果在SAR图像舰船数据集HRSID(high-resolution SAR images dataset)上的实验结果表明,本文方法应用在Cascade-RCNN(cascade region convolutional neural network)、YOLOv4(you only look once v4)和BorderDet(border detection)3种检测模型上,与原模型相比,对小型舰船的检测精度分别提高了3.5%、2.6%和2.9%,总体精度达到89.9%。在SSDD(SAR ship detection dataset)数据集上的总体精度达到95.9%,优于现有方法。结论本文通过改进骨干网络,使模型能够改变卷积核形状和大小,集中获取目标信息,抑制背景信息干扰,有效降低了SAR图像近岸复杂背景下小型舰船的误检漏检情况。
