针对人脸检测中小尺度人脸和遮挡人脸的漏检问题,提出了一种基于改进YOLOv5s-face(you only look once version 5 small-face)的Face5系列人脸检测算法Face5S(face5 small)和Face5M(face5 medium)。使用马赛克(mosaic)和图像混合(mixup...针对人脸检测中小尺度人脸和遮挡人脸的漏检问题,提出了一种基于改进YOLOv5s-face(you only look once version 5 small-face)的Face5系列人脸检测算法Face5S(face5 small)和Face5M(face5 medium)。使用马赛克(mosaic)和图像混合(mixup)数据增强方法,提升算法在复杂场景下检测人脸的泛化性和稳定性;通过改进C3的网络结构和引入可变形卷积(DCNv2)降低算法的参数量,提高算法提取特征的灵活性;通过引入特征的内容感知重组上采样算子(CARAFE),提高多尺度人脸的检测性能;引入损失函数WIoUV3(wise intersection over union version 3),提升算法的小尺度人脸检测性能。实验结果表明,在WIDER FACE验证集上,相较于YOLOv5s-face算法,Face5S算法的平均mAP@0.5提升了1.03%;相较于先进的人脸检测算法ASFD-D3(automatic and scalable face detector-D3)和TinaFace,Face5M算法的平均mAP@0.5分别提升了1.07%和2.11%,提出的Face5系列算法能够有效提升算法对小尺度和部分遮挡人脸的检测性能,同时具有实时性。展开更多
人脸识别技术广泛应用于考勤管理、移动支付等智慧建设中。伴随着常态化的口罩干扰,传统人脸识别算法已无法满足实际应用需求,为此,本文利用深度学习模型SSD以及FaceNet模型对人脸识别系统展开设计。首先,为消除现有数据集中亚洲人脸占...人脸识别技术广泛应用于考勤管理、移动支付等智慧建设中。伴随着常态化的口罩干扰,传统人脸识别算法已无法满足实际应用需求,为此,本文利用深度学习模型SSD以及FaceNet模型对人脸识别系统展开设计。首先,为消除现有数据集中亚洲人脸占比小造成的类内间距变化差距不明显的问题,在CAS-IA Web Face公开数据集的基础上对亚洲人脸数据进行扩充;其次,为解决不同口罩样式对特征提取的干扰,使用SSD人脸检测模型与DLIB人脸关键点检测模型提取人脸关键点,并利用人脸关键点与口罩的空间位置关系,额外随机生成不同的口罩人脸,组成混合数据集;最后,在混合数据集上进行模型训练并将训练好的模型移植到人脸识别系统中,进行检测速度与识别精度验证。实验结果表明,系统的实时识别速度达20 fps以上,人脸识别模型准确率在构建的混合数据集中达到97.1%,在随机抽取的部分LFW数据集验证的准确率达99.7%,故而该系统可满足实际应用需求,在一定程度上提高人脸识别的鲁棒性与准确性。展开更多
In this research, we study the relationship between mental workload and facial temperature of aircraft participants during a simulated takeoff flight. We conducted experiments to comprehend the correlation between wor...In this research, we study the relationship between mental workload and facial temperature of aircraft participants during a simulated takeoff flight. We conducted experiments to comprehend the correlation between work and facial temperature within the flight simulator. The experiment involved a group of 10 participants who played the role of pilots in a simulated A-320 flight. Six different flying scenarios were designed to simulate normal and emergency situations on airplane takeoff that would occur in different levels of mental workload for the participants. The measurements were workload assessment, face temperatures, and heart rate monitoring. Throughout the experiments, we collected a total of 120 instances of takeoffs, together with over 10 hours of time-series data including heart rate, workload, and face thermal images and temperatures. Comparative analysis of EEG data and thermal image types, revealed intriguing findings. The results indicate a notable inverse relationship between workload and facial muscle temperatures, as well as facial landmark points. The results of this study contribute to a deeper understanding of the physiological effects of workload, as well as practical implications for aviation safety and performance.展开更多
Despite cities being recognized as being potential sources of microplastic pollution to the wider environment, most surveys of COVID-19 plastic-based litter have been undertaken through linear transects of marine beac...Despite cities being recognized as being potential sources of microplastic pollution to the wider environment, most surveys of COVID-19 plastic-based litter have been undertaken through linear transects of marine beaches. For the far fewer number of studies conducted on inland and urban locations, the site-specific focus has primarily been surveys along the length of streets. The present study is the first to specifically assess the standing stock (i.e., moment-in-time) of littered face masks for the entire surface area of urban parking lots. The density of face masks in 50 parking lots in a Canadian coastal town (0.00054 m2 ± 0.00051 m2) was found to be significantly greater than the background level of littering of town streets. Face mask density was significantly related to visitation “usage” of parking lots as gauged by the areal size of the lots and of their onsite buildings, as well as the number of vehicles present. Neither parking lot typology nor estimates of inferred export (various measures of wind exposure) and entrapment (various metrics of obstruction) of face masks had a significant influence on the extent of whole-lot littering. In consequence, modelling of the potential input of mask-derived microplastics to the marine environment from coastal communities can use the areal density of face masks found here in association with the total surface area of lots for individual municipalities as determined through GIS analysis.展开更多
针对目前大规模网络不适合在手机、平板电脑等资源匮乏的移动设备上使用,以及池化层会导致特征图的稀疏性最终影响神经网络识别精度的问题,提出了一个轻量级人脸识别神经网络ShuffaceNet,设计了一个非线性平滑Log-Mean-Exp函数ThetaMEX...针对目前大规模网络不适合在手机、平板电脑等资源匮乏的移动设备上使用,以及池化层会导致特征图的稀疏性最终影响神经网络识别精度的问题,提出了一个轻量级人脸识别神经网络ShuffaceNet,设计了一个非线性平滑Log-Mean-Exp函数ThetaMEX,并提出了一种端到端可训练的ThetaMEX全局池化层(TGPL),从而在保证算法精度的前提下,减少网络参数、提高运算速度,进而达到有效地将该网络部署在资源匮乏的移动设备上的目的。ShuffaceNet约有3 600个参数,模型大小仅为3.5 MB。在LFW(Labled Faces in the Wild)、AgeDB-30 (Age Database-30)、CFP (Celebrities in Frontal Profile)人脸数据集上的识别测试的结果表明,ShuffaceNet的精度分别达到了99.32%、93.17%、94.51%。与MobileNetV1、SqueezeNet、Xception相比,所提网络的大小分别缩减了73.1%、82.1%、78.5%,在AgeDB-30数据集上的精度分别提高了5.0%、6.3%、6.7%。可见,基于ThetaMEX全局池化的所提网络能够提高模型精度。展开更多
A face-mask object detection model incorporating hybrid dilation convolutional network termed ResNet Hybrid-dilation-convolution Face-mask-detector (RHF) is proposed in this paper. Furthermore, a lightweight face-mask...A face-mask object detection model incorporating hybrid dilation convolutional network termed ResNet Hybrid-dilation-convolution Face-mask-detector (RHF) is proposed in this paper. Furthermore, a lightweight face-mask dataset named Light Masked Face Dataset (LMFD) and a medium-sized face-mask dataset named Masked Face Dataset (MFD) with data augmentation methods applied is also constructed in this paper. The hybrid dilation convolutional network is able to expand the perception of the convolutional kernel without concern about the discontinuity of image information during the convolution process. For the given two datasets being constructed above, the trained models are significantly optimized in terms of detection performance, training time, and other related metrics. By using the MFD dataset of 55,905 images, the RHF model requires roughly 10 hours less training time compared to ResNet50 with better detection results with mAP of 93.45%.展开更多
Leiomyosarcoma is a rare malignant tumour of the lower limbs. Its differential histological diagnosis is difficult and is made in the presence of young scar tissue, leimyoma, dermatofibroma, melanoma, rabdomyosarcoma,...Leiomyosarcoma is a rare malignant tumour of the lower limbs. Its differential histological diagnosis is difficult and is made in the presence of young scar tissue, leimyoma, dermatofibroma, melanoma, rabdomyosarcoma, sarcomatoid carcinoma, fibroxantoma, Darrier Ferrand dermatofibrosarcoma and myofibroblastic tumours. Treatment is essentially surgical, with margins of 3 to 5 centimetres. We report two observations of tumours localised to the face, including one case of a known leiomyosarcoma and another case initially diagnosed as a leiomyosarcoma which turned out to be a cellular myofibroma with no sign of malignancy after several readings. The aim of this work is to review the literature on this pathology while highlighting the diagnostic and therapeutic difficulties. Conclusion: A rare smooth muscle tumour with a high risk of local recurrence in the event of incomplete treatment, leiomyosarcoma in its dermal component is preferentially located in the head and neck. Its treatment is exclusively surgical and highly mutilating.展开更多
文摘针对人脸检测中小尺度人脸和遮挡人脸的漏检问题,提出了一种基于改进YOLOv5s-face(you only look once version 5 small-face)的Face5系列人脸检测算法Face5S(face5 small)和Face5M(face5 medium)。使用马赛克(mosaic)和图像混合(mixup)数据增强方法,提升算法在复杂场景下检测人脸的泛化性和稳定性;通过改进C3的网络结构和引入可变形卷积(DCNv2)降低算法的参数量,提高算法提取特征的灵活性;通过引入特征的内容感知重组上采样算子(CARAFE),提高多尺度人脸的检测性能;引入损失函数WIoUV3(wise intersection over union version 3),提升算法的小尺度人脸检测性能。实验结果表明,在WIDER FACE验证集上,相较于YOLOv5s-face算法,Face5S算法的平均mAP@0.5提升了1.03%;相较于先进的人脸检测算法ASFD-D3(automatic and scalable face detector-D3)和TinaFace,Face5M算法的平均mAP@0.5分别提升了1.07%和2.11%,提出的Face5系列算法能够有效提升算法对小尺度和部分遮挡人脸的检测性能,同时具有实时性。
文摘人脸识别技术广泛应用于考勤管理、移动支付等智慧建设中。伴随着常态化的口罩干扰,传统人脸识别算法已无法满足实际应用需求,为此,本文利用深度学习模型SSD以及FaceNet模型对人脸识别系统展开设计。首先,为消除现有数据集中亚洲人脸占比小造成的类内间距变化差距不明显的问题,在CAS-IA Web Face公开数据集的基础上对亚洲人脸数据进行扩充;其次,为解决不同口罩样式对特征提取的干扰,使用SSD人脸检测模型与DLIB人脸关键点检测模型提取人脸关键点,并利用人脸关键点与口罩的空间位置关系,额外随机生成不同的口罩人脸,组成混合数据集;最后,在混合数据集上进行模型训练并将训练好的模型移植到人脸识别系统中,进行检测速度与识别精度验证。实验结果表明,系统的实时识别速度达20 fps以上,人脸识别模型准确率在构建的混合数据集中达到97.1%,在随机抽取的部分LFW数据集验证的准确率达99.7%,故而该系统可满足实际应用需求,在一定程度上提高人脸识别的鲁棒性与准确性。
文摘In this research, we study the relationship between mental workload and facial temperature of aircraft participants during a simulated takeoff flight. We conducted experiments to comprehend the correlation between work and facial temperature within the flight simulator. The experiment involved a group of 10 participants who played the role of pilots in a simulated A-320 flight. Six different flying scenarios were designed to simulate normal and emergency situations on airplane takeoff that would occur in different levels of mental workload for the participants. The measurements were workload assessment, face temperatures, and heart rate monitoring. Throughout the experiments, we collected a total of 120 instances of takeoffs, together with over 10 hours of time-series data including heart rate, workload, and face thermal images and temperatures. Comparative analysis of EEG data and thermal image types, revealed intriguing findings. The results indicate a notable inverse relationship between workload and facial muscle temperatures, as well as facial landmark points. The results of this study contribute to a deeper understanding of the physiological effects of workload, as well as practical implications for aviation safety and performance.
文摘Despite cities being recognized as being potential sources of microplastic pollution to the wider environment, most surveys of COVID-19 plastic-based litter have been undertaken through linear transects of marine beaches. For the far fewer number of studies conducted on inland and urban locations, the site-specific focus has primarily been surveys along the length of streets. The present study is the first to specifically assess the standing stock (i.e., moment-in-time) of littered face masks for the entire surface area of urban parking lots. The density of face masks in 50 parking lots in a Canadian coastal town (0.00054 m2 ± 0.00051 m2) was found to be significantly greater than the background level of littering of town streets. Face mask density was significantly related to visitation “usage” of parking lots as gauged by the areal size of the lots and of their onsite buildings, as well as the number of vehicles present. Neither parking lot typology nor estimates of inferred export (various measures of wind exposure) and entrapment (various metrics of obstruction) of face masks had a significant influence on the extent of whole-lot littering. In consequence, modelling of the potential input of mask-derived microplastics to the marine environment from coastal communities can use the areal density of face masks found here in association with the total surface area of lots for individual municipalities as determined through GIS analysis.
文摘针对目前大规模网络不适合在手机、平板电脑等资源匮乏的移动设备上使用,以及池化层会导致特征图的稀疏性最终影响神经网络识别精度的问题,提出了一个轻量级人脸识别神经网络ShuffaceNet,设计了一个非线性平滑Log-Mean-Exp函数ThetaMEX,并提出了一种端到端可训练的ThetaMEX全局池化层(TGPL),从而在保证算法精度的前提下,减少网络参数、提高运算速度,进而达到有效地将该网络部署在资源匮乏的移动设备上的目的。ShuffaceNet约有3 600个参数,模型大小仅为3.5 MB。在LFW(Labled Faces in the Wild)、AgeDB-30 (Age Database-30)、CFP (Celebrities in Frontal Profile)人脸数据集上的识别测试的结果表明,ShuffaceNet的精度分别达到了99.32%、93.17%、94.51%。与MobileNetV1、SqueezeNet、Xception相比,所提网络的大小分别缩减了73.1%、82.1%、78.5%,在AgeDB-30数据集上的精度分别提高了5.0%、6.3%、6.7%。可见,基于ThetaMEX全局池化的所提网络能够提高模型精度。
文摘A face-mask object detection model incorporating hybrid dilation convolutional network termed ResNet Hybrid-dilation-convolution Face-mask-detector (RHF) is proposed in this paper. Furthermore, a lightweight face-mask dataset named Light Masked Face Dataset (LMFD) and a medium-sized face-mask dataset named Masked Face Dataset (MFD) with data augmentation methods applied is also constructed in this paper. The hybrid dilation convolutional network is able to expand the perception of the convolutional kernel without concern about the discontinuity of image information during the convolution process. For the given two datasets being constructed above, the trained models are significantly optimized in terms of detection performance, training time, and other related metrics. By using the MFD dataset of 55,905 images, the RHF model requires roughly 10 hours less training time compared to ResNet50 with better detection results with mAP of 93.45%.
文摘Leiomyosarcoma is a rare malignant tumour of the lower limbs. Its differential histological diagnosis is difficult and is made in the presence of young scar tissue, leimyoma, dermatofibroma, melanoma, rabdomyosarcoma, sarcomatoid carcinoma, fibroxantoma, Darrier Ferrand dermatofibrosarcoma and myofibroblastic tumours. Treatment is essentially surgical, with margins of 3 to 5 centimetres. We report two observations of tumours localised to the face, including one case of a known leiomyosarcoma and another case initially diagnosed as a leiomyosarcoma which turned out to be a cellular myofibroma with no sign of malignancy after several readings. The aim of this work is to review the literature on this pathology while highlighting the diagnostic and therapeutic difficulties. Conclusion: A rare smooth muscle tumour with a high risk of local recurrence in the event of incomplete treatment, leiomyosarcoma in its dermal component is preferentially located in the head and neck. Its treatment is exclusively surgical and highly mutilating.
