Daphniopsis tibetana is widely distributed in Xinjiang,Qinghai,Tibet of China,as well as in Russia and India,which is the dominant zooplankton in many high-altitude(4000 m)salt lakes.D.tibetana can adapt to saline wat...Daphniopsis tibetana is widely distributed in Xinjiang,Qinghai,Tibet of China,as well as in Russia and India,which is the dominant zooplankton in many high-altitude(4000 m)salt lakes.D.tibetana can adapt to saline waters,whereas the other species of the superorder Cladocera can only inhabit in freshwater.However,the phylogenetic status of D.tibetana in Branchiopoda remains unclear primarily because of limited mitogenome.In this study,complete mitochondrial genome sequences of D.tibetana were sequenced and annotated for the first time to obtain a comprehensive understanding of its phylogenetic status.The complete mitogenome of D.tibetana is 16196 bp in length.It contains 37 genes,including two ribosomal RNAs(12S and 16S rRNAs)genes,22 transfer RNA(tRNA)genes,13 protein-coding genes,and one non-coding region.The overall base composition is 29.6%A,33.2%T,19.0%G,and 18.2%C with a high AT bias(62.8%).Except for trnS1(GCT),most tRNAs have a typical cloverleaf secondary structure.Phylogenetic analysis based on maximum likelihood and Bayesian inference generates identical phylogenetic topology and shows the phylogenetic status of D.tibetana,which reconfirm the distinction between the genera Daphniopsis and Daphnia.Meanwhile,the class Branchiopoda is clustered into three clades(Anostraca,Notostraca,and Diplostraca)with high support values.These results provide not only a comprehensive understanding of the characteristics of D.tibetana mitogenome and its phylogenetic position in Diplostraca,but also information for future research on the phylogeny of Branchiopoda.展开更多
非结构化道路的缺陷目标检测任务对道路交通安全具有重要意义,但检测所需的标注数据集相对有限。为了解决非结构化道路标注数据集缺乏以及现有模型对无标注数据学习能力不足的问题,提出一种MAM(Multi-Augmentation with Memory)半监督...非结构化道路的缺陷目标检测任务对道路交通安全具有重要意义,但检测所需的标注数据集相对有限。为了解决非结构化道路标注数据集缺乏以及现有模型对无标注数据学习能力不足的问题,提出一种MAM(Multi-Augmentation with Memory)半监督目标检测算法。首先,引入缓存机制存储无标注图像和带有伪标注图像的框回归位置信息,避免了后续匹配造成的计算资源浪费。其次,设计混合数据增强策略,将缓存的伪标签图像与无标签图像混合输入学生模型,以增强模型对新数据的泛化能力,并使图像的尺度分布更加均衡。MAM算法不受目标检测模型的限制,并且更好地保持了目标框的一致性,避免了计算一致性损失。实验结果表明,MAM算法相比其他全监督学习和半监督学习算法更具优越性,在自建的非结构化道路缺陷数据集Defect上,在标注比例为10%、20%和30%的场景下,MAM算法的均值平均精度(mAP)相比于Soft Teacher算法分别提升了6.8、11.1和6.0百分点,在自建的非结构化道路坑洼数据集Pothole上,在标注比例为15%和30%的场景下,MAM算法的mAP相比于Soft Teacher算法分别提升了5.8和4.3百分点。展开更多
基金supported by the Zhejiang Provincial Natural Science Foundation of China(Nos.LY22D060001,LY20C190008)the National Natural Science Foundation of China(Nos.41806156,31702321)+3 种基金the Fund of Guangdong Provincial Key Laboratory of Fishery Ecology and Environment(FEEL-2021-8)the Open Foundation from Key Laboratory of Tropical Marine Bio-resources and Ecology,Chinese Academy of Sciences(No.LMB20201005)the Science and Technology Project of Zhoushan(No.2020 C21016),the Open Foundation from Marine Sciences in the First-Class Subjects of Zhejiang(Nos.20200201,20200202)the Starting Research Fund from the Zhejiang Ocean University.
文摘Daphniopsis tibetana is widely distributed in Xinjiang,Qinghai,Tibet of China,as well as in Russia and India,which is the dominant zooplankton in many high-altitude(4000 m)salt lakes.D.tibetana can adapt to saline waters,whereas the other species of the superorder Cladocera can only inhabit in freshwater.However,the phylogenetic status of D.tibetana in Branchiopoda remains unclear primarily because of limited mitogenome.In this study,complete mitochondrial genome sequences of D.tibetana were sequenced and annotated for the first time to obtain a comprehensive understanding of its phylogenetic status.The complete mitogenome of D.tibetana is 16196 bp in length.It contains 37 genes,including two ribosomal RNAs(12S and 16S rRNAs)genes,22 transfer RNA(tRNA)genes,13 protein-coding genes,and one non-coding region.The overall base composition is 29.6%A,33.2%T,19.0%G,and 18.2%C with a high AT bias(62.8%).Except for trnS1(GCT),most tRNAs have a typical cloverleaf secondary structure.Phylogenetic analysis based on maximum likelihood and Bayesian inference generates identical phylogenetic topology and shows the phylogenetic status of D.tibetana,which reconfirm the distinction between the genera Daphniopsis and Daphnia.Meanwhile,the class Branchiopoda is clustered into three clades(Anostraca,Notostraca,and Diplostraca)with high support values.These results provide not only a comprehensive understanding of the characteristics of D.tibetana mitogenome and its phylogenetic position in Diplostraca,but also information for future research on the phylogeny of Branchiopoda.
文摘非结构化道路的缺陷目标检测任务对道路交通安全具有重要意义,但检测所需的标注数据集相对有限。为了解决非结构化道路标注数据集缺乏以及现有模型对无标注数据学习能力不足的问题,提出一种MAM(Multi-Augmentation with Memory)半监督目标检测算法。首先,引入缓存机制存储无标注图像和带有伪标注图像的框回归位置信息,避免了后续匹配造成的计算资源浪费。其次,设计混合数据增强策略,将缓存的伪标签图像与无标签图像混合输入学生模型,以增强模型对新数据的泛化能力,并使图像的尺度分布更加均衡。MAM算法不受目标检测模型的限制,并且更好地保持了目标框的一致性,避免了计算一致性损失。实验结果表明,MAM算法相比其他全监督学习和半监督学习算法更具优越性,在自建的非结构化道路缺陷数据集Defect上,在标注比例为10%、20%和30%的场景下,MAM算法的均值平均精度(mAP)相比于Soft Teacher算法分别提升了6.8、11.1和6.0百分点,在自建的非结构化道路坑洼数据集Pothole上,在标注比例为15%和30%的场景下,MAM算法的mAP相比于Soft Teacher算法分别提升了5.8和4.3百分点。
