基于SSD-MobileNet的矿工安保穿戴设备检测系统
|
摘要
针对因矿山人员流动复杂性导致监管困难及因下井人员安全意识不强导致不能有效穿戴安保设备等问题,设计了基于SSD-MobileNet的矿工安保穿戴设备检测系统。将SSD算法的特征提取网络VGG16替换成MobileNet网络,构建了SSD-MobileNet算法模型;按照VOC2007数据集标准制作矿工安保穿戴设备的照片数据集,对SSD-MobileNet算法模型进行训练;采用SSD-MobileNet算法识别矿工8件安保穿戴设备(安全帽、防尘面具、工作服、工作靴、手电筒、自救器、定位卡、防砸背夹),根据矿工多张不同角度的照片的最高置信度结果综合判定是否穿戴了某件安保设备,并将最高置信度阈值设置为75%。测试结果表明,该系统能够实时、准确地检测出矿工安保设备的穿戴情况,且具有良好的稳定性和抗干扰性。
In view of problems of supervision difficulties caused by complexity of mine personnel flow,and that security equipments cannot be effectively worn due to weak safety awareness of underground personnel,a detection system of miners' wearable security equipments based on SSD-MobileNet was designed.Feature extraction network VGG16 of SSD algorithm is replaced by MobileNet network to construct SSD-MobileNet algorithm model.Photo data set of miners' wearable security equipments is made according to VOC2007 data set standard and used to train the SSD-MobileNet algorithm model.The SSDMobileNet algorithm is used to identify eight wearable security equipments for miners(hardhats,dust masks,overalls,work boots,flashlights,self-rescuing devices,positioning cards,anti-back clamps).The highest confidence of photos of miner with multiple angles is used for comprehensive determination of whether a security equipment is worn and the highest confidence threshold is set at 75%.The test results show that the system can detect wearing condition of miners' security equipments in real time and accurately,and has good stability and anti-interference.
In view of problems of supervision difficulties caused by complexity of mine personnel flow,and that security equipments cannot be effectively worn due to weak safety awareness of underground personnel,a detection system of miners' wearable security equipments based on SSD-MobileNet was designed.Feature extraction network VGG16 of SSD algorithm is replaced by MobileNet network to construct SSD-MobileNet algorithm model.Photo data set of miners' wearable security equipments is made according to VOC2007 data set standard and used to train the SSD-MobileNet algorithm model.The SSDMobileNet algorithm is used to identify eight wearable security equipments for miners(hardhats,dust masks,overalls,work boots,flashlights,self-rescuing devices,positioning cards,anti-back clamps).The highest confidence of photos of miner with multiple angles is used for comprehensive determination of whether a security equipment is worn and the highest confidence threshold is set at 75%.The test results show that the system can detect wearing condition of miners' security equipments in real time and accurately,and has good stability and anti-interference.
引文
[1]田仙仙,鲍泓,徐成.一种改进HOG特征的行人检测算法[J].计算机科学,2014,41(9):320-324.TIAN Xianxian,BAO Hong,XU Cheng.Improved HOG algorithm of pedestrian detection[J].Computer Science,2014,41(9):320-324.
[2]王庆伟,应自炉.一种基于Haar-Like T特征的人脸检测算法[J].模式识别与人工智能,2015,28(1):35-41.WANG Qingwei,YING Zilu.A face detection algorithm based on Haar-Like T features[J].Pattern Recognition and Artificial Intelligence,2015,28(1):35-41.
[3]何云,吴怀宇,钟锐.基于多种LBP特征集成学习的人脸识别[J].计算机应用研究,2018,35(1):292-295.HE Yun,WU Huaiyu,ZHONG Rui.Face recognition based on ensemble learning with multiple LBPfeatures[J].Application Research of Computers,2018,35(1):292-295.
[4]尹宝才,王文通,王立春.深度学习研究综述[J].北京工业大学学报,2015,41(1):48-59.YIN Baocai,WANG Wentong,WANG Lichun.Review of deep learning[J].Journal of Beijing University of Technology,2015,41(1):48-59.
[5]孙志远,鲁成祥,史忠植,等.深度学习研究与进展[J].计算机科学,2016,43(2):1-8.SUN Zhiyuan,LU Chengxiang,SHI Zhongzhi,et al.Research and advances on deep learning[J].Computer Science,2016,43(2):1-8.
[6]郑胤,陈权崎,章毓晋.深度学习及其在目标和行为识别中的新进展[J].中国图象图形学报,2014,19(2):175-184.ZHENG Yin,CHEN Quanqi,ZHANG Yujin.Deep learning and its new progress in object and behavior recognition[J].Journal of Image and Graphics,2014,19(2):175-184.
[7]段萌.基于卷积神经网络的图像识别方法研究[D].郑州:郑州大学,2017.
[8]卢宏涛,张秦川.深度卷积神经网络在计算机视觉中的应用研究综述[J].数据采集与处理,2016,31(1):1-17.LU Hongtao,ZHANG Qinchuan.Applications of deep convolutional neural network in computer vision[J].Journal of Data Acquisition and Processing,2016,31(1):1-17.
[9]唐士宇,朱艾春,张赛,等.基于深度卷积神经网络的井下人员目标检测[J].工矿自动化,2018,44(11):32-36.TANG Shiyu,ZHU Aichun,ZHANG Sai,et al.Target detection of underground personnel based on deep convolutional neural network[J].Industry and Mine Automation,2018,44(11):32-36.
[10]GIRSHICK R.Fast R-CNN[C]//IEEE International Conference on Computer Vision,Santiago,Chile,2015.
[11]REN S,HE K,GIRSHICK R,et al.Faster R-CNN:Towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2017,39(6):1137-1149.
[12]DAI J,LI Y,HE K,et al.R-FCN:Object detection via region-based fully convolutional networks[C]//Advances in Neural Information Processing Systems,2016.
[13]REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:unified,real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2016.
[14]LIU W,ANGUELOV D,ERHAN D,et al.SSD:Single shot multibox detector[C]//European Conference on Computer Vision,2016.
[2]王庆伟,应自炉.一种基于Haar-Like T特征的人脸检测算法[J].模式识别与人工智能,2015,28(1):35-41.WANG Qingwei,YING Zilu.A face detection algorithm based on Haar-Like T features[J].Pattern Recognition and Artificial Intelligence,2015,28(1):35-41.
[3]何云,吴怀宇,钟锐.基于多种LBP特征集成学习的人脸识别[J].计算机应用研究,2018,35(1):292-295.HE Yun,WU Huaiyu,ZHONG Rui.Face recognition based on ensemble learning with multiple LBPfeatures[J].Application Research of Computers,2018,35(1):292-295.
[4]尹宝才,王文通,王立春.深度学习研究综述[J].北京工业大学学报,2015,41(1):48-59.YIN Baocai,WANG Wentong,WANG Lichun.Review of deep learning[J].Journal of Beijing University of Technology,2015,41(1):48-59.
[5]孙志远,鲁成祥,史忠植,等.深度学习研究与进展[J].计算机科学,2016,43(2):1-8.SUN Zhiyuan,LU Chengxiang,SHI Zhongzhi,et al.Research and advances on deep learning[J].Computer Science,2016,43(2):1-8.
[6]郑胤,陈权崎,章毓晋.深度学习及其在目标和行为识别中的新进展[J].中国图象图形学报,2014,19(2):175-184.ZHENG Yin,CHEN Quanqi,ZHANG Yujin.Deep learning and its new progress in object and behavior recognition[J].Journal of Image and Graphics,2014,19(2):175-184.
[7]段萌.基于卷积神经网络的图像识别方法研究[D].郑州:郑州大学,2017.
[8]卢宏涛,张秦川.深度卷积神经网络在计算机视觉中的应用研究综述[J].数据采集与处理,2016,31(1):1-17.LU Hongtao,ZHANG Qinchuan.Applications of deep convolutional neural network in computer vision[J].Journal of Data Acquisition and Processing,2016,31(1):1-17.
[9]唐士宇,朱艾春,张赛,等.基于深度卷积神经网络的井下人员目标检测[J].工矿自动化,2018,44(11):32-36.TANG Shiyu,ZHU Aichun,ZHANG Sai,et al.Target detection of underground personnel based on deep convolutional neural network[J].Industry and Mine Automation,2018,44(11):32-36.
[10]GIRSHICK R.Fast R-CNN[C]//IEEE International Conference on Computer Vision,Santiago,Chile,2015.
[11]REN S,HE K,GIRSHICK R,et al.Faster R-CNN:Towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2017,39(6):1137-1149.
[12]DAI J,LI Y,HE K,et al.R-FCN:Object detection via region-based fully convolutional networks[C]//Advances in Neural Information Processing Systems,2016.
[13]REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:unified,real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2016.
[14]LIU W,ANGUELOV D,ERHAN D,et al.SSD:Single shot multibox detector[C]//European Conference on Computer Vision,2016.