Moirématerials,composed of two single-layer two-dimensional semiconductors,are important because they are good platforms for studying strongly correlated physics.Among them,moirématerials based on transition...Moirématerials,composed of two single-layer two-dimensional semiconductors,are important because they are good platforms for studying strongly correlated physics.Among them,moirématerials based on transition metal dichalcogenides(TMDs)have been intensively studied.The hetero-bilayer can support moiréinterlayer excitons if there is a small twist angle or small lattice constant difference between the TMDs in the hetero-bilayer and form a type-Ⅱ band alignment.The coupling of moiréinterlayer excitons to cavity modes can induce exotic phenomena.Here,we review recent advances in the coupling of moiréinterlayer excitons to cavities,and comment on the current difficulties and possible future research directions in this field.展开更多
车辆型号识别在智能交通系统、涉车刑侦案件侦破等方面具有十分重要的应用前景.针对车辆型号种类繁多、部分型号区分度小等带来的车辆型号精细分类困难的问题,采用车辆正脸图像为数据源,提出一种多分支多维度特征融合的卷积神经网络模型...车辆型号识别在智能交通系统、涉车刑侦案件侦破等方面具有十分重要的应用前景.针对车辆型号种类繁多、部分型号区分度小等带来的车辆型号精细分类困难的问题,采用车辆正脸图像为数据源,提出一种多分支多维度特征融合的卷积神经网络模型Fg-CarNet (Convolutional neural networks for car fine-grained classification, Fg-CarNet).该模型根据车正脸图像特征分布特点,将其分为上下两部分并行进行特征提取,并对网络中间层产生的特征进行两个维度的融合,以提取有区分度的特征,提高特征表达能力,通过使用小卷积核以及全局均值池化,使在网络分类准确度提高的同时降低了网络模型参数大小.在CompCars数据集上进行验证,实验结果表明, Fg-CarNet提取的车辆特征在保证网络模型参数最小的同时,车辆型号识别率达到最高,实现了最好的分类效果.展开更多
基金supported by the National Key R&D Program of China(Grant No.2018YFA036900)the Beijing Natural Science Foundation(Grant No.JQ21018)。
文摘Moirématerials,composed of two single-layer two-dimensional semiconductors,are important because they are good platforms for studying strongly correlated physics.Among them,moirématerials based on transition metal dichalcogenides(TMDs)have been intensively studied.The hetero-bilayer can support moiréinterlayer excitons if there is a small twist angle or small lattice constant difference between the TMDs in the hetero-bilayer and form a type-Ⅱ band alignment.The coupling of moiréinterlayer excitons to cavity modes can induce exotic phenomena.Here,we review recent advances in the coupling of moiréinterlayer excitons to cavities,and comment on the current difficulties and possible future research directions in this field.
文摘车辆型号识别在智能交通系统、涉车刑侦案件侦破等方面具有十分重要的应用前景.针对车辆型号种类繁多、部分型号区分度小等带来的车辆型号精细分类困难的问题,采用车辆正脸图像为数据源,提出一种多分支多维度特征融合的卷积神经网络模型Fg-CarNet (Convolutional neural networks for car fine-grained classification, Fg-CarNet).该模型根据车正脸图像特征分布特点,将其分为上下两部分并行进行特征提取,并对网络中间层产生的特征进行两个维度的融合,以提取有区分度的特征,提高特征表达能力,通过使用小卷积核以及全局均值池化,使在网络分类准确度提高的同时降低了网络模型参数大小.在CompCars数据集上进行验证,实验结果表明, Fg-CarNet提取的车辆特征在保证网络模型参数最小的同时,车辆型号识别率达到最高,实现了最好的分类效果.
