The Five-hundred-meter Aperture Spherical radio Telescope(FAST)is the world’s largest single-dish radio telescope.Its large reflecting surface achieves unprecedented sensitivity but is prone to damage,such as dents a...The Five-hundred-meter Aperture Spherical radio Telescope(FAST)is the world’s largest single-dish radio telescope.Its large reflecting surface achieves unprecedented sensitivity but is prone to damage,such as dents and holes,caused by naturally-occurring falling objects.Hence,the timely and accurate detection of surface defects is crucial for FAST’s stable operation.Conventional manual inspection involves human inspectors climbing up and examining the large surface visually,a time-consuming and potentially unreliable process.To accelerate the inspection process and increase its accuracy,this work makes the first step towards automating the inspection of FAST by integrating deep-learning techniques with drone technology.First,a drone flies over the surface along a predetermined route.Since surface defects significantly vary in scale and show high inter-class similarity,directly applying existing deep detectors to detect defects on the drone imagery is highly prone to missing and misidentifying defects.As a remedy,we introduce cross-fusion,a dedicated plug-in operation for deep detectors that enables the adaptive fusion of multi-level features in a point-wise selective fashion,depending on local defect patterns.Consequently,strong semantics and fine-grained details are dynamically fused at different positions to support the accurate detection of defects of various scales and types.Our AI-powered drone-based automated inspection is time-efficient,reliable,and has good accessibility,which guarantees the long-term and stable operation of FAST.展开更多
Agriculture plays an important role in the economy of all countries.However,plant diseases may badly affect the quality of food,production,and ultimately the economy.For plant disease detection and management,agricult...Agriculture plays an important role in the economy of all countries.However,plant diseases may badly affect the quality of food,production,and ultimately the economy.For plant disease detection and management,agriculturalists spend a huge amount of money.However,the manual detection method of plant diseases is complicated and time-consuming.Consequently,automated systems for plant disease detection using machine learning(ML)approaches are proposed.However,most of the existing ML techniques of plants diseases recognition are based on handcrafted features and they rarely deal with huge amount of input data.To address the issue,this article proposes a fully automated method for plant disease detection and recognition using deep neural networks.In the proposed method,AlexNet and VGG19 CNNs are considered as pre-trained architectures.It is capable to obtain the feature extraction of the given data with fine-tuning details.After convolutional neural network feature extraction,it selects the best subset of features through the correlation coefficient and feeds them to the number of classifiers including K-Nearest Neighbor,Support Vector Machine,Probabilistic Neural Network,Fuzzy logic,and Artificial Neural Network.The validation of the proposed method is carried out on a self-collected dataset generated through the augmentation step.The achieved average accuracy of our method is more than 96%and outperforms the recent techniques.展开更多
多功能视频编码(versatile video coding,VVC)是最新的视频编码标准,与高效视频编码(high efficiency video coding,HEVC)相比进一步提高了压缩效率,但由于引入了包括二叉树和三叉树在内的多类树结构,同时帧内角度模式从35种增加到67种...多功能视频编码(versatile video coding,VVC)是最新的视频编码标准,与高效视频编码(high efficiency video coding,HEVC)相比进一步提高了压缩效率,但由于引入了包括二叉树和三叉树在内的多类树结构,同时帧内角度模式从35种增加到67种,导致编码复杂度剧增。为了降低计算复杂度,本文提出了一种基于快速编码单元(coding unit,CU)划分和角度模式决策的VVC帧内编码算法。首先根据自适应标准差阈值对CU纹理复杂度进行分类,初步缩减划分模式列表;然后采用Sobel梯度算子确定纹理方向,跳过非最优划分模式;最后根据统计结果筛选淘汰掉概率小于2%的角度模式。实验结果表明,与VTM-2.1相比,该算法能节省51.05%的编码时间,BDBR(Bjontegarrd delta bit rate)仅上升1.98%。展开更多
基金financially supported by the National Natural Science Foundation of China(No.62101032)the Postdoctoral Science Foundation of China(Nos.2021M690015,2022T150050)the Beijing Institute of Technology Research Fund Program for Young Scholars(No.3040011182111).
文摘The Five-hundred-meter Aperture Spherical radio Telescope(FAST)is the world’s largest single-dish radio telescope.Its large reflecting surface achieves unprecedented sensitivity but is prone to damage,such as dents and holes,caused by naturally-occurring falling objects.Hence,the timely and accurate detection of surface defects is crucial for FAST’s stable operation.Conventional manual inspection involves human inspectors climbing up and examining the large surface visually,a time-consuming and potentially unreliable process.To accelerate the inspection process and increase its accuracy,this work makes the first step towards automating the inspection of FAST by integrating deep-learning techniques with drone technology.First,a drone flies over the surface along a predetermined route.Since surface defects significantly vary in scale and show high inter-class similarity,directly applying existing deep detectors to detect defects on the drone imagery is highly prone to missing and misidentifying defects.As a remedy,we introduce cross-fusion,a dedicated plug-in operation for deep detectors that enables the adaptive fusion of multi-level features in a point-wise selective fashion,depending on local defect patterns.Consequently,strong semantics and fine-grained details are dynamically fused at different positions to support the accurate detection of defects of various scales and types.Our AI-powered drone-based automated inspection is time-efficient,reliable,and has good accessibility,which guarantees the long-term and stable operation of FAST.
基金the MSIT(Ministry of Science and ICT),Korea,under the ITRC(Information Technology Research Center)support program(IITP-2020-2016-0-00312)supervised by the IITP(Institute for Information&Communications Technology Planning&Evaluation)in part by the MSIP(Ministry of Science,ICT&Future Planning),Korea,under the National Program for Excellence in SW)(2015-0-00938)supervised by the IITP(Institute for Information&communications Technology Planning&Evaluation).
文摘Agriculture plays an important role in the economy of all countries.However,plant diseases may badly affect the quality of food,production,and ultimately the economy.For plant disease detection and management,agriculturalists spend a huge amount of money.However,the manual detection method of plant diseases is complicated and time-consuming.Consequently,automated systems for plant disease detection using machine learning(ML)approaches are proposed.However,most of the existing ML techniques of plants diseases recognition are based on handcrafted features and they rarely deal with huge amount of input data.To address the issue,this article proposes a fully automated method for plant disease detection and recognition using deep neural networks.In the proposed method,AlexNet and VGG19 CNNs are considered as pre-trained architectures.It is capable to obtain the feature extraction of the given data with fine-tuning details.After convolutional neural network feature extraction,it selects the best subset of features through the correlation coefficient and feeds them to the number of classifiers including K-Nearest Neighbor,Support Vector Machine,Probabilistic Neural Network,Fuzzy logic,and Artificial Neural Network.The validation of the proposed method is carried out on a self-collected dataset generated through the augmentation step.The achieved average accuracy of our method is more than 96%and outperforms the recent techniques.
文摘多功能视频编码(versatile video coding,VVC)是最新的视频编码标准,与高效视频编码(high efficiency video coding,HEVC)相比进一步提高了压缩效率,但由于引入了包括二叉树和三叉树在内的多类树结构,同时帧内角度模式从35种增加到67种,导致编码复杂度剧增。为了降低计算复杂度,本文提出了一种基于快速编码单元(coding unit,CU)划分和角度模式决策的VVC帧内编码算法。首先根据自适应标准差阈值对CU纹理复杂度进行分类,初步缩减划分模式列表;然后采用Sobel梯度算子确定纹理方向,跳过非最优划分模式;最后根据统计结果筛选淘汰掉概率小于2%的角度模式。实验结果表明,与VTM-2.1相比,该算法能节省51.05%的编码时间,BDBR(Bjontegarrd delta bit rate)仅上升1.98%。
