Due to the consideration of safety,non-contact measurement methods are be-coming more acceptable.However,massive measurement will bring high labor-cost and low working efficiency.To address these limitations,this pape...Due to the consideration of safety,non-contact measurement methods are be-coming more acceptable.However,massive measurement will bring high labor-cost and low working efficiency.To address these limitations,this paper introduces a deep learning model for the antenna attitude parameter measurement,which can be divided into an an-tenna location phase and a calculation phase of the attitude parameter.In the first phase,a single shot multibox detector(SSD)is applied to automatically recognize and discover the antenna from pictures taken by drones.In the second phase,the located antennas’fea-ture lines are extracted and their attitude parameters are then calculated mathematically.Experiments show that the proposed algorithms outperform existing related works in effi-ciency and accuracy,and therefore can be effectively used in engineering applications.展开更多
We propose an improved a single-shot detector(SSD)algorithm to detect falls of the elderly.The VGG16 network part of the SSD network is replaced with the MobilenetV2 network.At the same time,we change the infrastructu...We propose an improved a single-shot detector(SSD)algorithm to detect falls of the elderly.The VGG16 network part of the SSD network is replaced with the MobilenetV2 network.At the same time,we change the infrastructure of MobilenetV2 network,the three layers that were not down-sampled at the end were removed,which can make the model structure simpler and faster to detect.The complete Intersection-over-Union(CIoU)loss function is introduced to get a good regression of the target borders that have different sizes and different proportions.We use Feature Pyramid Network(FPN)for up-sampling,it can fuse low-level feature maps with high resolution and high-level feature maps with rich semantic information.For sampling results,we use the Secure Shell(SSH)module to extract different receptive fields,which improves the detection accuracy.Our model ensures that the accuracy of the elderly fall detection remains unchanged,but it greatly improves the detection speed that only takes 10 milliseconds to detect a picture.展开更多
文摘Due to the consideration of safety,non-contact measurement methods are be-coming more acceptable.However,massive measurement will bring high labor-cost and low working efficiency.To address these limitations,this paper introduces a deep learning model for the antenna attitude parameter measurement,which can be divided into an an-tenna location phase and a calculation phase of the attitude parameter.In the first phase,a single shot multibox detector(SSD)is applied to automatically recognize and discover the antenna from pictures taken by drones.In the second phase,the located antennas’fea-ture lines are extracted and their attitude parameters are then calculated mathematically.Experiments show that the proposed algorithms outperform existing related works in effi-ciency and accuracy,and therefore can be effectively used in engineering applications.
基金supported by the National Natural Science Foundation of China under Grant No.61572038the Innovation Project Foundation NCUT.
文摘We propose an improved a single-shot detector(SSD)algorithm to detect falls of the elderly.The VGG16 network part of the SSD network is replaced with the MobilenetV2 network.At the same time,we change the infrastructure of MobilenetV2 network,the three layers that were not down-sampled at the end were removed,which can make the model structure simpler and faster to detect.The complete Intersection-over-Union(CIoU)loss function is introduced to get a good regression of the target borders that have different sizes and different proportions.We use Feature Pyramid Network(FPN)for up-sampling,it can fuse low-level feature maps with high resolution and high-level feature maps with rich semantic information.For sampling results,we use the Secure Shell(SSH)module to extract different receptive fields,which improves the detection accuracy.Our model ensures that the accuracy of the elderly fall detection remains unchanged,but it greatly improves the detection speed that only takes 10 milliseconds to detect a picture.
