针对当前船舶检测模型复杂度高、对设备资源要求较高等问题,提出一种基于YOLOv5s的船舶目标检测算法SRE-YOLOv5s.该算法使用ShuffleNetV2轻量级网络替换YOLOv5s原始特征提取主干网络降低模型复杂度,并使用感受野模块(Receptive Field Bl...针对当前船舶检测模型复杂度高、对设备资源要求较高等问题,提出一种基于YOLOv5s的船舶目标检测算法SRE-YOLOv5s.该算法使用ShuffleNetV2轻量级网络替换YOLOv5s原始特征提取主干网络降低模型复杂度,并使用感受野模块(Receptive Field Block,RFB)增强算法特征处理能力,考虑到Complete IoU(CIoU)损失函数会影响模型收敛速度,采用Efficient IoU(EIoU)损失函数进行优化.通过对SRE-YOLOv5s算法进行训练和验证,结果表明:SRE-YOLOv5s在轻量化的同时保证了检测精度,与其他主流轻量化检测模型相比,SRE-YOLOv5s具有更好的检测性能.此外,根据船舶可视化检测结果可知,SRE-YOLOv5s模型在实际场景中具有较大的应用潜力.展开更多
AIM:To characterize the distribution of meibomian gland(MG)area loss(MGL)and its relationship with demographic characteristics,mites,and symptoms.METHODS:This retrospective observational study included patients who vi...AIM:To characterize the distribution of meibomian gland(MG)area loss(MGL)and its relationship with demographic characteristics,mites,and symptoms.METHODS:This retrospective observational study included patients who visited the Dry Eye Clinic of Shenzhen Eye Hospital between June 2020 and August 2021.General patient characteristics,ocular symptoms,Demodex test results of the eyelid edges,and the results of a comprehensive ocular surface analysis were collected.MGL was analyzed using Image J software.RESULTS:This study enrolled 1204 outpatients aged 20-80(40.70±13.44)y,including 357 males(29.65%)and 847 females(70.35%).The patients were classified into mild(n=155;12.87%),moderate(n=795;66.03%),severe(n=206;17.11%),and extremely severe(n=48;3.99%)MGL groups.MGL was significantly larger in female than in male(P=0.006).The degree of MGL also significantly differed in age(P<0.001)and the more numbers of mites with severity(P<0.001).Multivariate disordered multinomial logistic regression analysis identified that female sex,older age,secretory symptoms,and a large number of mites were risk factors for MGL(P<0.05).CONCLUSION:Patients with MGL are more likely to be older,female,more numbers of mites,and increased secretion.展开更多
文摘针对当前船舶检测模型复杂度高、对设备资源要求较高等问题,提出一种基于YOLOv5s的船舶目标检测算法SRE-YOLOv5s.该算法使用ShuffleNetV2轻量级网络替换YOLOv5s原始特征提取主干网络降低模型复杂度,并使用感受野模块(Receptive Field Block,RFB)增强算法特征处理能力,考虑到Complete IoU(CIoU)损失函数会影响模型收敛速度,采用Efficient IoU(EIoU)损失函数进行优化.通过对SRE-YOLOv5s算法进行训练和验证,结果表明:SRE-YOLOv5s在轻量化的同时保证了检测精度,与其他主流轻量化检测模型相比,SRE-YOLOv5s具有更好的检测性能.此外,根据船舶可视化检测结果可知,SRE-YOLOv5s模型在实际场景中具有较大的应用潜力.
基金Supported by the Science,Technology,and Innovation Commission of Shenzhen Municipality(JCYJ20230807114605011).
文摘AIM:To characterize the distribution of meibomian gland(MG)area loss(MGL)and its relationship with demographic characteristics,mites,and symptoms.METHODS:This retrospective observational study included patients who visited the Dry Eye Clinic of Shenzhen Eye Hospital between June 2020 and August 2021.General patient characteristics,ocular symptoms,Demodex test results of the eyelid edges,and the results of a comprehensive ocular surface analysis were collected.MGL was analyzed using Image J software.RESULTS:This study enrolled 1204 outpatients aged 20-80(40.70±13.44)y,including 357 males(29.65%)and 847 females(70.35%).The patients were classified into mild(n=155;12.87%),moderate(n=795;66.03%),severe(n=206;17.11%),and extremely severe(n=48;3.99%)MGL groups.MGL was significantly larger in female than in male(P=0.006).The degree of MGL also significantly differed in age(P<0.001)and the more numbers of mites with severity(P<0.001).Multivariate disordered multinomial logistic regression analysis identified that female sex,older age,secretory symptoms,and a large number of mites were risk factors for MGL(P<0.05).CONCLUSION:Patients with MGL are more likely to be older,female,more numbers of mites,and increased secretion.
