Climate change presents a critical global challenge,threatening human well-being,ecosystems,economies,and societies.While mitigation efforts remain essential and critically important,the growing urgency of climate imp...Climate change presents a critical global challenge,threatening human well-being,ecosystems,economies,and societies.While mitigation efforts remain essential and critically important,the growing urgency of climate impacts necessitates immediate and effective adaptation measures.Effective adaptation strategies require advanced modeling tools with higher resolution,integration of ecosystem and social dynamics,and the ability to assess diverse adaptation scenarios.Local-scale models,which are performed at the scale of an administrative region,a country,or a specified region,are particularly valuable as they can incorporate specific adaptation measures and generate precise,contextspecific insights.These models play a key role in formulating tailored climate adaptation strategies and action plans.This paper explores the significance and challenges in developing such models,emphasizing the pressing need to accelerate their advancement.We call on the scientific community and policymakers to prioritize the development of tailored local-scale modeling tools and services to enhance resilience and better support adaptive responses to the complex and evolving challenges posed by climate change and rapid urbanization at the local level.展开更多
堆煤是输送机常见故障之一,为了保障煤矿工业生产的安全,需要对煤矿井下输送机的堆煤情况进行检测。然而现有的检测方法存在容易误触、检测可靠性较差等缺点,针对这些问题提出一种基于Transformer统一多尺度时序卷积(unified multi-scal...堆煤是输送机常见故障之一,为了保障煤矿工业生产的安全,需要对煤矿井下输送机的堆煤情况进行检测。然而现有的检测方法存在容易误触、检测可靠性较差等缺点,针对这些问题提出一种基于Transformer统一多尺度时序卷积(unified multi-scale temporal ConvTransformer,UnMS-TCT)网络用于输送机堆煤检测。首先融合RGB帧和光流帧提取的特征,使网络更全面地建模时空关系;然后在时序编码器中,将动态位置嵌入(dynamic position embedding,DPE),多头关系聚合器(multi-head relation aggregator,MHRA)以及多层感知机(multilayer perceptron,MLP)组成的全局模块,交叉注意力(cross-attention,CA)组成的局部模块,以交替方式形成全局-局部关系模块,增强多尺度下获取全局和局部时间关系的能力;其次利用残差全局-局部融合(residual global and local fusion,ResGLFus)模块融合多尺度特征,有效地提高融合过程的稳定性,最终实现高精度堆煤预测。实验结果表明:该方法能够实现对输送机堆煤的检测,mAP达到98.17%。展开更多
针对火焰检测过程中存在小目标难以检测的问题,提出了一种改进的YOLOv8n模型。首先,在双分支跨阶段局部特征融合(cross stage partial 2 with feature fusion,C2f)模块中加入动态蛇形卷积,有助于提取多尺度特征、增强特征表示。接着,将G...针对火焰检测过程中存在小目标难以检测的问题,提出了一种改进的YOLOv8n模型。首先,在双分支跨阶段局部特征融合(cross stage partial 2 with feature fusion,C2f)模块中加入动态蛇形卷积,有助于提取多尺度特征、增强特征表示。接着,将GhostnetV2引入到颈部网络中,不仅减少了模型的参数量,还提升了整体的检测精度和速度。然后,添加微小目标检测头以便更好地进行多尺度小目标的检测,基于局部和全局的挤压激励(squeeze and excitation,SE)注意力机制确保每一层的特征都得到充分优化,特别是小目标的细微特征。最后,基于最小点距离的交并比损失函数提高算法的收敛速度和定位精度。实验结果显示,改进YOLOv8n模型的P、R、FPS、mAP@0.5和mAP@0.5∶0.95指标平均值比YOLOv8n模型分别提高了3.34%、3.62%、14帧/s、3.01%和3.41%,表明模型拥有较好的小目标火焰检测能力。研究结果可为预防火灾等安全事故提供理论依据和决策支撑。展开更多
基金support of the National Natural Science Foundation of China(Nos.42288101&42375183)Shanghai International Science and Technology Partnership Project(No.21230780200)+1 种基金Shanghai B&R Joint Laboratory Project(No.22230750300)EU HORIZON Project FOCI(No.101056783).
文摘Climate change presents a critical global challenge,threatening human well-being,ecosystems,economies,and societies.While mitigation efforts remain essential and critically important,the growing urgency of climate impacts necessitates immediate and effective adaptation measures.Effective adaptation strategies require advanced modeling tools with higher resolution,integration of ecosystem and social dynamics,and the ability to assess diverse adaptation scenarios.Local-scale models,which are performed at the scale of an administrative region,a country,or a specified region,are particularly valuable as they can incorporate specific adaptation measures and generate precise,contextspecific insights.These models play a key role in formulating tailored climate adaptation strategies and action plans.This paper explores the significance and challenges in developing such models,emphasizing the pressing need to accelerate their advancement.We call on the scientific community and policymakers to prioritize the development of tailored local-scale modeling tools and services to enhance resilience and better support adaptive responses to the complex and evolving challenges posed by climate change and rapid urbanization at the local level.
文摘堆煤是输送机常见故障之一,为了保障煤矿工业生产的安全,需要对煤矿井下输送机的堆煤情况进行检测。然而现有的检测方法存在容易误触、检测可靠性较差等缺点,针对这些问题提出一种基于Transformer统一多尺度时序卷积(unified multi-scale temporal ConvTransformer,UnMS-TCT)网络用于输送机堆煤检测。首先融合RGB帧和光流帧提取的特征,使网络更全面地建模时空关系;然后在时序编码器中,将动态位置嵌入(dynamic position embedding,DPE),多头关系聚合器(multi-head relation aggregator,MHRA)以及多层感知机(multilayer perceptron,MLP)组成的全局模块,交叉注意力(cross-attention,CA)组成的局部模块,以交替方式形成全局-局部关系模块,增强多尺度下获取全局和局部时间关系的能力;其次利用残差全局-局部融合(residual global and local fusion,ResGLFus)模块融合多尺度特征,有效地提高融合过程的稳定性,最终实现高精度堆煤预测。实验结果表明:该方法能够实现对输送机堆煤的检测,mAP达到98.17%。
文摘针对火焰检测过程中存在小目标难以检测的问题,提出了一种改进的YOLOv8n模型。首先,在双分支跨阶段局部特征融合(cross stage partial 2 with feature fusion,C2f)模块中加入动态蛇形卷积,有助于提取多尺度特征、增强特征表示。接着,将GhostnetV2引入到颈部网络中,不仅减少了模型的参数量,还提升了整体的检测精度和速度。然后,添加微小目标检测头以便更好地进行多尺度小目标的检测,基于局部和全局的挤压激励(squeeze and excitation,SE)注意力机制确保每一层的特征都得到充分优化,特别是小目标的细微特征。最后,基于最小点距离的交并比损失函数提高算法的收敛速度和定位精度。实验结果显示,改进YOLOv8n模型的P、R、FPS、mAP@0.5和mAP@0.5∶0.95指标平均值比YOLOv8n模型分别提高了3.34%、3.62%、14帧/s、3.01%和3.41%,表明模型拥有较好的小目标火焰检测能力。研究结果可为预防火灾等安全事故提供理论依据和决策支撑。
