The hardness, elastic modulus, nano-scratch resistance and wear depth for a bulk metallic glass of and its partial crystallization alloys have been measured by using nanoindentation method. The results showed that par...The hardness, elastic modulus, nano-scratch resistance and wear depth for a bulk metallic glass of and its partial crystallization alloys have been measured by using nanoindentation method. The results showed that partial crystallization did not influence the reduced elastic modulus but increased the hardness, and then increased the scratch coefficient. The scratch coefficient increased linearly with increasing the hardness H but decreases when W>6.2GPa. Partial crystallization decreased evidently the wear depth, and when the load was large the wear depth decreased with increasing the hardness.展开更多
为缓解YOLOv7在检测个人防护用品时面临标签重写、标签分配不平衡和特征耦合等问题,提出一种基于改进YOLOv7的检测方法.首先去除YOLOv7的大尺度和中尺度输出层,以降低标签重写率,且保证输出层得到充分训练;其次将输出层的定位和分类解耦...为缓解YOLOv7在检测个人防护用品时面临标签重写、标签分配不平衡和特征耦合等问题,提出一种基于改进YOLOv7的检测方法.首先去除YOLOv7的大尺度和中尺度输出层,以降低标签重写率,且保证输出层得到充分训练;其次将输出层的定位和分类解耦,避免不同任务的特征表示互相影响,并选择在边界框级别检测防护服,在关键点级别检测防护帽和防护手套;最后引入部分卷积,实现实时检测.为验证该方法的有效性,使用实验人员穿戴防护用品的图像数据对所提方法进行验证.结果表明,相比YOLOv7,该方法的精确率和召回率分别提高了4.1和4.5个百分点,FPS(Frames Per Second)提升了1.3帧,可满足实验室场景下的个人防护用品穿戴检测需求.展开更多
基金support by the special funds for the major state basic research(2002AA331080).
文摘The hardness, elastic modulus, nano-scratch resistance and wear depth for a bulk metallic glass of and its partial crystallization alloys have been measured by using nanoindentation method. The results showed that partial crystallization did not influence the reduced elastic modulus but increased the hardness, and then increased the scratch coefficient. The scratch coefficient increased linearly with increasing the hardness H but decreases when W>6.2GPa. Partial crystallization decreased evidently the wear depth, and when the load was large the wear depth decreased with increasing the hardness.
文摘为缓解YOLOv7在检测个人防护用品时面临标签重写、标签分配不平衡和特征耦合等问题,提出一种基于改进YOLOv7的检测方法.首先去除YOLOv7的大尺度和中尺度输出层,以降低标签重写率,且保证输出层得到充分训练;其次将输出层的定位和分类解耦,避免不同任务的特征表示互相影响,并选择在边界框级别检测防护服,在关键点级别检测防护帽和防护手套;最后引入部分卷积,实现实时检测.为验证该方法的有效性,使用实验人员穿戴防护用品的图像数据对所提方法进行验证.结果表明,相比YOLOv7,该方法的精确率和召回率分别提高了4.1和4.5个百分点,FPS(Frames Per Second)提升了1.3帧,可满足实验室场景下的个人防护用品穿戴检测需求.
