It is well known that [6,6]-phenyl-C<sub><span style="font-size:12px;font-family:Verdana;">61</span></sub><span style="font-size:12px;font-family:Verdana;">-butyric ac...It is well known that [6,6]-phenyl-C<sub><span style="font-size:12px;font-family:Verdana;">61</span></sub><span style="font-size:12px;font-family:Verdana;">-butyric acid methyl ester (PCBM) is a common n-type passivation material in PSCs, usually used as an interface modification layer. However, PCBM is extremely expensive and is not suitable for future industrialization. Herein, the various concentrations of PCBM as an additive are adopted for PSCs. It not only avoids the routine process of spin coating the multi-layer films, but also reduces the PCBM material and cost. Meanwhile, PCBM can passivate the grain surface and modulate morphology of perovskite films. Furthermore, the most important optical parameters of solar cells, the current density (</span><i><span style="font-size:12px;font-family:Verdana;">J</span><sub><span style="font-size:12px;font-family:Verdana;">sc</span></sub></i><span style="font-size:12px;font-family:Verdana;">), fill factor (FF), open-circuit voltage (</span><i><span style="font-size:12px;font-family:Verdana;">V</span><sub><span style="font-size:12px;font-family:Verdana;">oc</span></sub></i><span style="font-size:12px;font-family:Verdana;">) and power conversion efficiencies (PCE) were improved. Especially, when the PCBM doping ratio in CH</span><sub><span style="font-size:12px;font-family:Verdana;">3</span></sub><span style="font-size:12px;font-family:Verdana;">NH</span><sub><span style="font-size:12px;font-family:Verdana;">3</span></sub><span style="font-size:12px;font-family:Verdana;">PbI</span><sub><span style="font-size:12px;font-family:Verdana;">3</span></sub><span style="font-size:12px;font-family:Verdana;"> (MAPbI</span><sub><span style="font-size:12px;font-family:Verdana;">3</span></sub><span style="font-size:12px;font-family:Verdana;">) precursor solution was 1</span><span><span><span style="font-family:;" "=""> </span></span></span><span><span><span style="font-family:;" "=""><span style="font-size:12px;font-family:Verdana;">wt%, the device obtained the smallest </span><i><span style="font-size:12px;font-family:Verdana;">V</span><sub><span style="font-size:12px;font-family:Verdana;">oc</span></sub></i><span style="font-size:12px;font-family:Verdana;"> decay (less than 1%) in the p-i-n type PSCs with poly</span></span></span></span><span><span><span style="font-family:;" "=""> </span></span></span><span><span><span style="font-family:;" "=""><span style="font-size:12px;font-family:Verdana;">(3,4-ethylenedioxythiophene):poly (styrene sulfonate) (PEDOT:PSS) as hole transport layer (HTL) and fullerene (C</span><sub><span style="font-size:12px;font-family:Verdana;">60</span></sub><span style="font-size:12px;font-family:Verdana;">) as electron transport layer (ETL). The PSCs </span><i><span style="font-size:12px;font-family:Verdana;">V</span><sub><span style="font-size:12px;font-family:Verdana;">oc</span></sub></i><span style="font-size:12px;font-family:Verdana;"> stability improvement is attri</span><span style="font-size:12px;font-family:Verdana;">buted to enhanced crystallinity of photoactive layer and decreased non-radiative </span><span style="font-size:12px;font-family:Verdana;">recombination by PCBM doping in the perovskites.</span></span></span></span>展开更多
针对无人机航拍图像目标检测中视野变化大、时空信息复杂等问题,文中基于YOLOv5(You Only Look Once Version5)架构,提出基于图像低维特征融合的航拍小目标检测模型.引入CA(Coordinate Attention),改进MobileNetV3的反转残差块,增加图...针对无人机航拍图像目标检测中视野变化大、时空信息复杂等问题,文中基于YOLOv5(You Only Look Once Version5)架构,提出基于图像低维特征融合的航拍小目标检测模型.引入CA(Coordinate Attention),改进MobileNetV3的反转残差块,增加图像空间维度信息的同时降低模型参数量.改进YOLOv5特征金字塔网络结构,融合浅层网络中的特征图,增加模型对图像低维有效信息的表达能力,进而提升小目标检测精度.同时为了降低航拍图像中复杂背景带来的干扰,引入无参平均注意力模块,同时关注图像的空间注意力与通道注意力;引入VariFocal Loss,降低负样本在训练过程中的权重占比.在VisDrone数据集上的实验验证文中模型的有效性,该模型在有效提升检测精度的同时明显降低复杂度.展开更多
随着深度学习在国内目标检测的不断应用,常规的大、中目标检测已经取得惊人的进步,但由于卷积网络本身的局限性,针对小目标检测依然会出现漏检、误检的问题,以数据集Visdrone2019和数据集FloW-Img为例,对YOLOv7模型进行研究,在网络结构...随着深度学习在国内目标检测的不断应用,常规的大、中目标检测已经取得惊人的进步,但由于卷积网络本身的局限性,针对小目标检测依然会出现漏检、误检的问题,以数据集Visdrone2019和数据集FloW-Img为例,对YOLOv7模型进行研究,在网络结构上对骨干网的ELAN模块进行改进,将Focal NeXt block加入到ELAN模块的长短梯度路径中融合来强化输出小目标的特征质量和提高输出特征包含的上下文信息含量,在头部网络引入RepLKDeXt模块,该模块不仅可以取代SPPCSPC模块来简化模型整体结构还可以利用多通道、大卷积核和Cat操作来优化ELAN-H结构,最后引入SIOU损失函数取代CIOU函数以此提高该模型的鲁棒性。结果表明改进后的YOLOv7模型参数量减少计算复杂性降低并在小目标密度高的Visdrone 2019数据集上的检测性能近似不变,在小目标稀疏的FloW-Img数据集上涨幅9.05个百分点,进一步简化了模型并增加了模型的适用范围。展开更多
针对无人机视角下的小目标检测精度较差、漏检较为严重的问题,提出一种基于改进YOLOv5的无人机图像检测算法。针对小目标尺度较小问题在骨干网络替换空间金字塔池化(Spatial Pyramid Pooling,SPP)为SPPCSPC-GS,增强密集区域关注能力,提...针对无人机视角下的小目标检测精度较差、漏检较为严重的问题,提出一种基于改进YOLOv5的无人机图像检测算法。针对小目标尺度较小问题在骨干网络替换空间金字塔池化(Spatial Pyramid Pooling,SPP)为SPPCSPC-GS,增强密集区域关注能力,提取更多小目标有效特征;在颈部网络中引入CBAM注意力机制将头部C3模块替换为C3CBAM增强上下文信息,提高空间与通道特征表达能力;针对遮挡问题引入柔性非极大值抑制(Soft Non Maximum Suppression,Soft NMS)提升模型对遮挡和密集目标的检测能力;替换损失函数为EIOU加快收敛提升定位效果。改进后的模型在VisDrone数据集上平均检测精度为42.2%,相较于原始YOLOv5s算法提升10.7%,遮挡严重的小目标行人与人类别精度分别上升12%与13.3%。相较于其他先进算法,所提算法表现优秀,可以满足无人机视角图像检测任务要求。展开更多
文摘It is well known that [6,6]-phenyl-C<sub><span style="font-size:12px;font-family:Verdana;">61</span></sub><span style="font-size:12px;font-family:Verdana;">-butyric acid methyl ester (PCBM) is a common n-type passivation material in PSCs, usually used as an interface modification layer. However, PCBM is extremely expensive and is not suitable for future industrialization. Herein, the various concentrations of PCBM as an additive are adopted for PSCs. It not only avoids the routine process of spin coating the multi-layer films, but also reduces the PCBM material and cost. Meanwhile, PCBM can passivate the grain surface and modulate morphology of perovskite films. Furthermore, the most important optical parameters of solar cells, the current density (</span><i><span style="font-size:12px;font-family:Verdana;">J</span><sub><span style="font-size:12px;font-family:Verdana;">sc</span></sub></i><span style="font-size:12px;font-family:Verdana;">), fill factor (FF), open-circuit voltage (</span><i><span style="font-size:12px;font-family:Verdana;">V</span><sub><span style="font-size:12px;font-family:Verdana;">oc</span></sub></i><span style="font-size:12px;font-family:Verdana;">) and power conversion efficiencies (PCE) were improved. Especially, when the PCBM doping ratio in CH</span><sub><span style="font-size:12px;font-family:Verdana;">3</span></sub><span style="font-size:12px;font-family:Verdana;">NH</span><sub><span style="font-size:12px;font-family:Verdana;">3</span></sub><span style="font-size:12px;font-family:Verdana;">PbI</span><sub><span style="font-size:12px;font-family:Verdana;">3</span></sub><span style="font-size:12px;font-family:Verdana;"> (MAPbI</span><sub><span style="font-size:12px;font-family:Verdana;">3</span></sub><span style="font-size:12px;font-family:Verdana;">) precursor solution was 1</span><span><span><span style="font-family:;" "=""> </span></span></span><span><span><span style="font-family:;" "=""><span style="font-size:12px;font-family:Verdana;">wt%, the device obtained the smallest </span><i><span style="font-size:12px;font-family:Verdana;">V</span><sub><span style="font-size:12px;font-family:Verdana;">oc</span></sub></i><span style="font-size:12px;font-family:Verdana;"> decay (less than 1%) in the p-i-n type PSCs with poly</span></span></span></span><span><span><span style="font-family:;" "=""> </span></span></span><span><span><span style="font-family:;" "=""><span style="font-size:12px;font-family:Verdana;">(3,4-ethylenedioxythiophene):poly (styrene sulfonate) (PEDOT:PSS) as hole transport layer (HTL) and fullerene (C</span><sub><span style="font-size:12px;font-family:Verdana;">60</span></sub><span style="font-size:12px;font-family:Verdana;">) as electron transport layer (ETL). The PSCs </span><i><span style="font-size:12px;font-family:Verdana;">V</span><sub><span style="font-size:12px;font-family:Verdana;">oc</span></sub></i><span style="font-size:12px;font-family:Verdana;"> stability improvement is attri</span><span style="font-size:12px;font-family:Verdana;">buted to enhanced crystallinity of photoactive layer and decreased non-radiative </span><span style="font-size:12px;font-family:Verdana;">recombination by PCBM doping in the perovskites.</span></span></span></span>
文摘针对无人机航拍图像目标检测中视野变化大、时空信息复杂等问题,文中基于YOLOv5(You Only Look Once Version5)架构,提出基于图像低维特征融合的航拍小目标检测模型.引入CA(Coordinate Attention),改进MobileNetV3的反转残差块,增加图像空间维度信息的同时降低模型参数量.改进YOLOv5特征金字塔网络结构,融合浅层网络中的特征图,增加模型对图像低维有效信息的表达能力,进而提升小目标检测精度.同时为了降低航拍图像中复杂背景带来的干扰,引入无参平均注意力模块,同时关注图像的空间注意力与通道注意力;引入VariFocal Loss,降低负样本在训练过程中的权重占比.在VisDrone数据集上的实验验证文中模型的有效性,该模型在有效提升检测精度的同时明显降低复杂度.
文摘随着深度学习在国内目标检测的不断应用,常规的大、中目标检测已经取得惊人的进步,但由于卷积网络本身的局限性,针对小目标检测依然会出现漏检、误检的问题,以数据集Visdrone2019和数据集FloW-Img为例,对YOLOv7模型进行研究,在网络结构上对骨干网的ELAN模块进行改进,将Focal NeXt block加入到ELAN模块的长短梯度路径中融合来强化输出小目标的特征质量和提高输出特征包含的上下文信息含量,在头部网络引入RepLKDeXt模块,该模块不仅可以取代SPPCSPC模块来简化模型整体结构还可以利用多通道、大卷积核和Cat操作来优化ELAN-H结构,最后引入SIOU损失函数取代CIOU函数以此提高该模型的鲁棒性。结果表明改进后的YOLOv7模型参数量减少计算复杂性降低并在小目标密度高的Visdrone 2019数据集上的检测性能近似不变,在小目标稀疏的FloW-Img数据集上涨幅9.05个百分点,进一步简化了模型并增加了模型的适用范围。
文摘针对无人机视角下的小目标检测精度较差、漏检较为严重的问题,提出一种基于改进YOLOv5的无人机图像检测算法。针对小目标尺度较小问题在骨干网络替换空间金字塔池化(Spatial Pyramid Pooling,SPP)为SPPCSPC-GS,增强密集区域关注能力,提取更多小目标有效特征;在颈部网络中引入CBAM注意力机制将头部C3模块替换为C3CBAM增强上下文信息,提高空间与通道特征表达能力;针对遮挡问题引入柔性非极大值抑制(Soft Non Maximum Suppression,Soft NMS)提升模型对遮挡和密集目标的检测能力;替换损失函数为EIOU加快收敛提升定位效果。改进后的模型在VisDrone数据集上平均检测精度为42.2%,相较于原始YOLOv5s算法提升10.7%,遮挡严重的小目标行人与人类别精度分别上升12%与13.3%。相较于其他先进算法,所提算法表现优秀,可以满足无人机视角图像检测任务要求。