To achieve radar and infrared stealth, an infrared stealth layer is usually added to the radar absorbing material（RAM） of stealth aircraft. By analyzing the millimeter-wave（MMW） emissivities of three stealth mater...To achieve radar and infrared stealth, an infrared stealth layer is usually added to the radar absorbing material（RAM） of stealth aircraft. By analyzing the millimeter-wave（MMW） emissivities of three stealth materials, this Letter investigates the impact of the added infrared stealth layer on the originally ＂hot＂ MMW emission of RAM. The theoretical and measured results indicate that, compared with the monolayer RAM, the MMW emission of the bilayer material is still strong and its emissivity is reduced by 0.1–0.2 at almost every incident angle.The results partially demonstrate the feasibility of detecting stealth aircraft coated with this bilayer stealth material.展开更多
Human pose estimation from image and video is a key task in many multimedia applications.Previous methods achieve great performance but rarely take efficiency into consideration,which makes it difficult to implement t...Human pose estimation from image and video is a key task in many multimedia applications.Previous methods achieve great performance but rarely take efficiency into consideration,which makes it difficult to implement the networks on lightweight devices.Nowadays,real-time multimedia applications call for more efficient models for better interaction.Moreover,most deep neural networks for pose estimation directly reuse networks designed for image classification as the backbone,which are not optimized for the pose estimation task.In this paper,we propose an efficient framework for human pose estimation with two parts,an efficient backbone and an efficient head.By implementing a differentiable neural architecture search method,we customize the backbone network design for pose estimation,and reduce computational cost with negligible accuracy degradation.For the efficient head,we slim the transposed convolutions and propose a spatial information correction module to promote the performance of the final prediction.In experiments,we evaluate our networks on the MPII and COCO datasets.Our smallest model requires only0.65 GFLOPs with 88.1%PCKh@0.5 on MPII and our large model needs only 2 GFLOPs while its accuracy is competitive with the state-of-the-art large model,HRNet,which takes 9.5 GFLOPs.展开更多
基金supported partially by the Shanghai Spaceflight Technology Renovation Fund(No.SAST2015088)the Fundamental Research Fund for the Central Universities(No.HUST2015QN093)
文摘To achieve radar and infrared stealth, an infrared stealth layer is usually added to the radar absorbing material（RAM） of stealth aircraft. By analyzing the millimeter-wave（MMW） emissivities of three stealth materials, this Letter investigates the impact of the added infrared stealth layer on the originally ＂hot＂ MMW emission of RAM. The theoretical and measured results indicate that, compared with the monolayer RAM, the MMW emission of the bilayer material is still strong and its emissivity is reduced by 0.1–0.2 at almost every incident angle.The results partially demonstrate the feasibility of detecting stealth aircraft coated with this bilayer stealth material.
基金supported by National Natural Science Foundation of China(NSFC)(Nos.61733007 and 61876212)Zhejiang Lab(No.2019NB0AB02)。
文摘Human pose estimation from image and video is a key task in many multimedia applications.Previous methods achieve great performance but rarely take efficiency into consideration,which makes it difficult to implement the networks on lightweight devices.Nowadays,real-time multimedia applications call for more efficient models for better interaction.Moreover,most deep neural networks for pose estimation directly reuse networks designed for image classification as the backbone,which are not optimized for the pose estimation task.In this paper,we propose an efficient framework for human pose estimation with two parts,an efficient backbone and an efficient head.By implementing a differentiable neural architecture search method,we customize the backbone network design for pose estimation,and reduce computational cost with negligible accuracy degradation.For the efficient head,we slim the transposed convolutions and propose a spatial information correction module to promote the performance of the final prediction.In experiments,we evaluate our networks on the MPII and COCO datasets.Our smallest model requires only0.65 GFLOPs with 88.1%PCKh@0.5 on MPII and our large model needs only 2 GFLOPs while its accuracy is competitive with the state-of-the-art large model,HRNet,which takes 9.5 GFLOPs.
