In the rapidly evolving urban landscape,outdoor parking lots have become an indispensable part of the city’s transportation system.The growth of parking lots has raised the likelihood of spontaneous vehicle combus-ti...In the rapidly evolving urban landscape,outdoor parking lots have become an indispensable part of the city’s transportation system.The growth of parking lots has raised the likelihood of spontaneous vehicle combus-tion,a significant safety hazard,making smoke detection an essential preventative step.However,the complex environment of outdoor parking lots presents additional challenges for smoke detection,which necessitates the development of more advanced and reliable smoke detection technologies.This paper addresses this concern and presents a novel smoke detection technique designed for the demanding environment of outdoor parking lots.First,we develop a novel dataset to fill the gap,as there is a lack of publicly available data.This dataset encompasses a wide range of smoke and fire scenarios,enhanced with data augmentation to ensure robustness against diverse outdoor conditions.Second,we utilize an optimized YOLOv5s model,integrated with the Squeeze-and-Excitation Network(SENet)attention mechanism,to significantly improve detection accuracy while maintaining real-time processing capabilities.Third,this paper implements an outdoor smoke detection system that is capable of accurately localizing and alerting in real time,enhancing the effectiveness and reliability of emergency response.Experiments show that the system has a high accuracy in terms of detecting smoke incidents in outdoor scenarios.展开更多
Measurements of carbon dioxide(CO_(2)),methane(CH_(4)),and carbon monoxide(CO)are of great importance in the Qinghai-Tibetan region,as it is the highest and largest plateau in the world affecting global weather and cl...Measurements of carbon dioxide(CO_(2)),methane(CH_(4)),and carbon monoxide(CO)are of great importance in the Qinghai-Tibetan region,as it is the highest and largest plateau in the world affecting global weather and climate systems.In this study,for the first time,we present CO_(2),CH_(4),and CO column measurements carried out by a Bruker EM27/SUN Fourier-transform infrared spectrometer(FTIR)at Golmud(36.42°E,94.91°N,2808 m)in August 2021.The mean and standard deviation of the column-average dry-air mixing ratio of CO_(2),CH_(4),and CO(XCO_(2),XCH_(4),and XCO)are 409.3±0.4 ppm,1905.5±19.4 ppb,and 103.1±7.7 ppb,respectively.The differences between the FTIR co-located TROPOMI/S5P satellite measurements at Golmud are 0.68±0.64%(13.1±12.2 ppb)for XCH_(4) and 9.81±3.48%(–10.7±3.8 ppb)for XCO,which are within their retrieval uncertainties.High correlations for both XCH_(4) and XCO are observed between the FTIR and S5P satellite measurements.Using the FLEXPART model and satellite measurements,we find that enhanced CH_(4) and CO columns in Golmud are affected by anthropogenic emissions transported from North India.This study provides an insight into the variations of the CO_(2),CH_(4),and CO columns in the Qinghai-Tibetan Plateau.展开更多
The quality of the stator winding coil directly affects the performance of the motor.A dual-camera online machine vision detection method to detect whether the coil leads and winding regions were qualified was designe...The quality of the stator winding coil directly affects the performance of the motor.A dual-camera online machine vision detection method to detect whether the coil leads and winding regions were qualified was designed.A vision detection platform was designed to capture individual winding images,and an image processing algorithm was used for image pre-processing,template matching and positioning of the coil lead area to set up a coordinate system.After eliminating image noise by Blob analysis,the improved Canny algorithm was used to detect the location of the coil lead paint stripped region,and the time was reduced by about half compared to the Canny algorithm.The coil winding region was trained with the ShuffleNet V2-YOLOv5s model for the dataset,and the detect file was converted to the Open Neural Network Exchange(ONNX)model for the detection of winding cross features with an average accuracy of 99.0%.The software interface of the detection system was designed to perform qualified discrimination tests on the workpieces,and the detection data were recorded and statistically analyzed.The results showed that the stator winding coil qualified discrimination accuracy reached 96.2%,and the average detection time of a single workpiece was about 300 ms,while YOLOv5s took less than 30 ms.展开更多
针对煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂工况环境因素导致煤矸识别存在误检、漏检以及检测精度低的问题,提出一种基于CFS-YOLO算法的煤矸智能识别模型。采用ConvNeXt V2(Convolutional Neural Network with NeXt Units...针对煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂工况环境因素导致煤矸识别存在误检、漏检以及检测精度低的问题,提出一种基于CFS-YOLO算法的煤矸智能识别模型。采用ConvNeXt V2(Convolutional Neural Network with NeXt Units Version 2)特征提取模块替换主干网络末端的2个C3(Cross Stage Partial Bottle Neck Mudule)模块,通过将掩码自动编码器(Masked Autoencoders,MAE)和全局响应归一化(Global Response Normalization,GRN)层添加到ConvNeXt架构中,有效缓解特征崩溃问题以及保持特征在网络传递过程中的多样性;采用Focal-EIOU(Focal and Efficient Intersection Over Union)损失函数替换原CIOU(Computer Intersection Over Union)损失函数,通过其Focal-Loss机制和调整样本权重的方式优化边界框回归任务中的样本不平衡问题,提高模型的收敛速度和定位精度;添加无参注意力机制(Simple Attention Mechanism,SimAM)于主干网络每个C3模块的后端,凭借其注意力权重自适应调整策略,提升模型对尺度变化较大或低分辨率煤矸目标关键特征的提取能力。通过消融试验和对比试验验证所提CFS-YOLO模型的有效性与优越性。试验结果表明:CFS-YOLO模型对于煤矸在煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂环境下的检测效果均得到有效提高,模型的平均精度均值达到90.2%,相较于原YOLOv5s模型的平均精度均值提高了3.7%,平均检测速度达到90.09 FPS,可充分满足煤矸实时检测的需求。同时与YOLOv5s、YOLOv7-tiny与YOLOv8n等6种YOLO系列算法相比,CFS-YOLO模型对煤矿复杂环境的适应性最强且综合检测性能最佳,可为煤矸的智能高效分选提供技术支持。展开更多
基金This work was supported byNatural Science Foundation of China(No.62362008,author Z.Z,https://www.nsfc.gov.cn/)Guizhou Provincial Science and Technology Projects(No.ZK[2022]149,author Z.Z,https://kjt.guizhou.gov.cn/)+2 种基金Guizhou Provincial Research Project(Youth)for Universities(No.[2022]104,author Z.Z,https://jyt.guizhou.gov.cn/)Natural Science Special Foundation of Guizhou University(No.[2021]47,author Z.Z,https://www.gzu.edu.cn/)GZU Cultivation Project of NSFC(No.[2020]80,author Z.Z,https://www.gzu.edu.cn/).
文摘In the rapidly evolving urban landscape,outdoor parking lots have become an indispensable part of the city’s transportation system.The growth of parking lots has raised the likelihood of spontaneous vehicle combus-tion,a significant safety hazard,making smoke detection an essential preventative step.However,the complex environment of outdoor parking lots presents additional challenges for smoke detection,which necessitates the development of more advanced and reliable smoke detection technologies.This paper addresses this concern and presents a novel smoke detection technique designed for the demanding environment of outdoor parking lots.First,we develop a novel dataset to fill the gap,as there is a lack of publicly available data.This dataset encompasses a wide range of smoke and fire scenarios,enhanced with data augmentation to ensure robustness against diverse outdoor conditions.Second,we utilize an optimized YOLOv5s model,integrated with the Squeeze-and-Excitation Network(SENet)attention mechanism,to significantly improve detection accuracy while maintaining real-time processing capabilities.Third,this paper implements an outdoor smoke detection system that is capable of accurately localizing and alerting in real time,enhancing the effectiveness and reliability of emergency response.Experiments show that the system has a high accuracy in terms of detecting smoke incidents in outdoor scenarios.
基金supported by the National Natural Science Foundation of China(Grant No.42205140,41975035)the National Key Research and Development Program of China(2021YFB3901000).
文摘Measurements of carbon dioxide(CO_(2)),methane(CH_(4)),and carbon monoxide(CO)are of great importance in the Qinghai-Tibetan region,as it is the highest and largest plateau in the world affecting global weather and climate systems.In this study,for the first time,we present CO_(2),CH_(4),and CO column measurements carried out by a Bruker EM27/SUN Fourier-transform infrared spectrometer(FTIR)at Golmud(36.42°E,94.91°N,2808 m)in August 2021.The mean and standard deviation of the column-average dry-air mixing ratio of CO_(2),CH_(4),and CO(XCO_(2),XCH_(4),and XCO)are 409.3±0.4 ppm,1905.5±19.4 ppb,and 103.1±7.7 ppb,respectively.The differences between the FTIR co-located TROPOMI/S5P satellite measurements at Golmud are 0.68±0.64%(13.1±12.2 ppb)for XCH_(4) and 9.81±3.48%(–10.7±3.8 ppb)for XCO,which are within their retrieval uncertainties.High correlations for both XCH_(4) and XCO are observed between the FTIR and S5P satellite measurements.Using the FLEXPART model and satellite measurements,we find that enhanced CH_(4) and CO columns in Golmud are affected by anthropogenic emissions transported from North India.This study provides an insight into the variations of the CO_(2),CH_(4),and CO columns in the Qinghai-Tibetan Plateau.
基金National Natural Science Foundation of China(No.U1831123)。
文摘The quality of the stator winding coil directly affects the performance of the motor.A dual-camera online machine vision detection method to detect whether the coil leads and winding regions were qualified was designed.A vision detection platform was designed to capture individual winding images,and an image processing algorithm was used for image pre-processing,template matching and positioning of the coil lead area to set up a coordinate system.After eliminating image noise by Blob analysis,the improved Canny algorithm was used to detect the location of the coil lead paint stripped region,and the time was reduced by about half compared to the Canny algorithm.The coil winding region was trained with the ShuffleNet V2-YOLOv5s model for the dataset,and the detect file was converted to the Open Neural Network Exchange(ONNX)model for the detection of winding cross features with an average accuracy of 99.0%.The software interface of the detection system was designed to perform qualified discrimination tests on the workpieces,and the detection data were recorded and statistically analyzed.The results showed that the stator winding coil qualified discrimination accuracy reached 96.2%,and the average detection time of a single workpiece was about 300 ms,while YOLOv5s took less than 30 ms.
文摘针对煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂工况环境因素导致煤矸识别存在误检、漏检以及检测精度低的问题,提出一种基于CFS-YOLO算法的煤矸智能识别模型。采用ConvNeXt V2(Convolutional Neural Network with NeXt Units Version 2)特征提取模块替换主干网络末端的2个C3(Cross Stage Partial Bottle Neck Mudule)模块,通过将掩码自动编码器(Masked Autoencoders,MAE)和全局响应归一化(Global Response Normalization,GRN)层添加到ConvNeXt架构中,有效缓解特征崩溃问题以及保持特征在网络传递过程中的多样性;采用Focal-EIOU(Focal and Efficient Intersection Over Union)损失函数替换原CIOU(Computer Intersection Over Union)损失函数,通过其Focal-Loss机制和调整样本权重的方式优化边界框回归任务中的样本不平衡问题,提高模型的收敛速度和定位精度;添加无参注意力机制(Simple Attention Mechanism,SimAM)于主干网络每个C3模块的后端,凭借其注意力权重自适应调整策略,提升模型对尺度变化较大或低分辨率煤矸目标关键特征的提取能力。通过消融试验和对比试验验证所提CFS-YOLO模型的有效性与优越性。试验结果表明:CFS-YOLO模型对于煤矸在煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂环境下的检测效果均得到有效提高,模型的平均精度均值达到90.2%,相较于原YOLOv5s模型的平均精度均值提高了3.7%,平均检测速度达到90.09 FPS,可充分满足煤矸实时检测的需求。同时与YOLOv5s、YOLOv7-tiny与YOLOv8n等6种YOLO系列算法相比,CFS-YOLO模型对煤矿复杂环境的适应性最强且综合检测性能最佳,可为煤矸的智能高效分选提供技术支持。
