The effect of the face thickness to core height ratio on different multi-layer pyramidal core sandwich columns under in-plane compression is investigated theoretically and numerically. Numerical simulation is in good ...The effect of the face thickness to core height ratio on different multi-layer pyramidal core sandwich columns under in-plane compression is investigated theoretically and numerically. Numerical simulation is in good agreement with theory. Results indicate that one specified face thickness to core height ratio corresponds to one optimum layer number of multi-layer pyramidal core sandwich columns in consideration of engineering application. This result can guide the sandwich structure design.展开更多
发电厂厂区内违规吸烟易导致火灾、爆炸等事故,会带来巨大损失;针对电厂内人员违规吸烟行为检测精度不高的问题,提出一种基于改进YOLOv5s(You Only Look Once v5s)的电厂内人员违规吸烟检测方法;该方法以YOLOv5s网络为基础,将YOLOv5s网...发电厂厂区内违规吸烟易导致火灾、爆炸等事故,会带来巨大损失;针对电厂内人员违规吸烟行为检测精度不高的问题,提出一种基于改进YOLOv5s(You Only Look Once v5s)的电厂内人员违规吸烟检测方法;该方法以YOLOv5s网络为基础,将YOLOv5s网络C3模块Bottleneck中的3×3卷积替换为多头自注意力层以提高算法的学习能力;接着在网络中添加ECA(Efficient Channel Attention)注意力模块,让网络更加关注待检测目标;同时将YOLOv5s网络的损失函数替换为SIoU(Scylla Intersection over Union),进一步提高算法的检测精度;最后采用加权双向特征金字塔网络(BiFPN,Bidirectional Feature Pyramid Network)代替原先YOLOv5s的特征金字塔网络,快速进行多尺度特征融合;实验结果表明,改进后算法吸烟行为的检测精度为89.3%,与改进前算法相比平均精度均值(mAP,mean Average Precision)提高了2.2%,检测效果显著提升,具有较高应用价值。展开更多
目标检测广泛应用在公共场合的智能监控、自动驾驶与计算机辅助诊断等领域。文章提出了单层特征目标检测替代复杂的特征金字塔结构,从而提升模型的推理速度和预测精度。在模型搭建过程中,瓶颈特征结构采用了单层空洞残差编码器,样本选...目标检测广泛应用在公共场合的智能监控、自动驾驶与计算机辅助诊断等领域。文章提出了单层特征目标检测替代复杂的特征金字塔结构,从而提升模型的推理速度和预测精度。在模型搭建过程中,瓶颈特征结构采用了单层空洞残差编码器,样本选择采用了统一匹配机制,并采用了任务对齐检测器。在COCO(Microsoft Common Objects in Context)数据集下,大量实验证明该方法的有效性,以Res Net50为基准,预测精度达到了38.2 m AP,比Retina Net的推理速度快1.4倍,精度提高2.3 m AP。该模型具有推理速度快、预测精度高等特点,可以应用在许多特定场景中。展开更多
基金supported by the National Natural Science Foundation of China under Grant No. 11432004
文摘The effect of the face thickness to core height ratio on different multi-layer pyramidal core sandwich columns under in-plane compression is investigated theoretically and numerically. Numerical simulation is in good agreement with theory. Results indicate that one specified face thickness to core height ratio corresponds to one optimum layer number of multi-layer pyramidal core sandwich columns in consideration of engineering application. This result can guide the sandwich structure design.
文摘发电厂厂区内违规吸烟易导致火灾、爆炸等事故,会带来巨大损失;针对电厂内人员违规吸烟行为检测精度不高的问题,提出一种基于改进YOLOv5s(You Only Look Once v5s)的电厂内人员违规吸烟检测方法;该方法以YOLOv5s网络为基础,将YOLOv5s网络C3模块Bottleneck中的3×3卷积替换为多头自注意力层以提高算法的学习能力;接着在网络中添加ECA(Efficient Channel Attention)注意力模块,让网络更加关注待检测目标;同时将YOLOv5s网络的损失函数替换为SIoU(Scylla Intersection over Union),进一步提高算法的检测精度;最后采用加权双向特征金字塔网络(BiFPN,Bidirectional Feature Pyramid Network)代替原先YOLOv5s的特征金字塔网络,快速进行多尺度特征融合;实验结果表明,改进后算法吸烟行为的检测精度为89.3%,与改进前算法相比平均精度均值(mAP,mean Average Precision)提高了2.2%,检测效果显著提升,具有较高应用价值。
文摘目标检测广泛应用在公共场合的智能监控、自动驾驶与计算机辅助诊断等领域。文章提出了单层特征目标检测替代复杂的特征金字塔结构,从而提升模型的推理速度和预测精度。在模型搭建过程中,瓶颈特征结构采用了单层空洞残差编码器,样本选择采用了统一匹配机制,并采用了任务对齐检测器。在COCO(Microsoft Common Objects in Context)数据集下,大量实验证明该方法的有效性,以Res Net50为基准,预测精度达到了38.2 m AP,比Retina Net的推理速度快1.4倍,精度提高2.3 m AP。该模型具有推理速度快、预测精度高等特点,可以应用在许多特定场景中。