It is difficult to differentiate small,but harmful,shell fragments of Chinese hickory nuts from their kernels since they are very similar in color.Including shell fragments of Chinese hickory nuts by mistake may creat...It is difficult to differentiate small,but harmful,shell fragments of Chinese hickory nuts from their kernels since they are very similar in color.Including shell fragments of Chinese hickory nuts by mistake may create safety hazards for consumers.Therefore,there is a need to develop an effective method to differentiate the shells from the kernels of Chinese hickory nuts.In this study,a deep learning approach based on a two-dimensional convolutional neural network(2D CNN)and long short-term memory(LSTM)integrated with hyperspectral imaging for distinguishing the shells and kernels of Chinese hickory nuts at the pixel level was proposed.Two classical classification methods,principal component analysis-K-nearest neighbors(PCA-KNN)and the support vector machine(SVM),were employed to establish identification models for comparison.The results showed that the 2D CNN-LSTM model achieved the best performance with an overall classification accuracy of 99.0%.Moreover,the shells in mixtures of shells and kernels were detected based on the proposed deep learning method and visualized for subsequent operations for the removal of foreign bodies.展开更多
Global cropland monitoring is important when considering tactical strategies for achieving food sustainability. Different global land cover (GLC) datasets providing cropland information have already been published and...Global cropland monitoring is important when considering tactical strategies for achieving food sustainability. Different global land cover (GLC) datasets providing cropland information have already been published and they are used in many applications. The different data input methods, classification techniques, class definitions and production years among the different GLC datasets make them all independently useful sources of information. This study attempted to produce a cropland agreement level (CAL) analysis based on the integration of several cropland datasets to more accurately estimate cropland area distribution. Estimating cropland area and how it has changed on a national level was done by converting the level of cropland agreement into percentages with an existing cropland fraction map. A pre-analysis showed that the four GLC datasets used in the 2005 and 2010 groups had similar year input data acquisitions. Therefore, we placed these four datasets (GlobCover, MODIS LC, GLCNMO and ESACCI LC) into 2005 and 2010 year-groups and selected them to process dataset integration through a CRISP approach. The results of this process proposed four agreement levels for this CAL analysis, and the model correlation was converted into percentage values. The cropland estimate results from the CAL analysis were observed along with FAO data statistics and showed the highest accuracy, with a 0.70 and 0.71 regression value for 2005 and 2010 respectively. In the cropland area change analysis, this CAL change analysis had the highest level of accuracy when describing the total size of cropland area change from 2005 and 2010 when compared to other individual original GLC datasets.展开更多
现有的异常检测方法能在特定应用场景下实现高精度检测,然而这些方法难以适用于其他应用场景,且自动化程度有限。因此,提出一种视觉基础模型(VFM)驱动的像素级图像异常检测方法SSMOD-Net(State Space Model driven-Omni Dimensional Ne...现有的异常检测方法能在特定应用场景下实现高精度检测,然而这些方法难以适用于其他应用场景,且自动化程度有限。因此,提出一种视觉基础模型(VFM)驱动的像素级图像异常检测方法SSMOD-Net(State Space Model driven-Omni Dimensional Net),旨在实现更精确的工业缺陷检测。与现有方法不同,SSMOD-Net实现SAM(Segment Anything Model)的自动化提示且不需要微调SAM,因此特别适用于需要处理大规模工业视觉数据的场景。SSMOD-Net的核心是一个新颖的提示编码器,该编码器由状态空间模型驱动,能够根据SAM的输入图像动态地生成提示。这一设计允许模型在保持SAM架构不变的同时,通过提示编码器引入额外的指导信息,从而提高检测精度。提示编码器内部集成一个残差多尺度模块,该模块基于状态空间模型构建,能够综合利用多尺度信息和全局信息。这一模块通过迭代搜索,在提示空间中寻找最优的提示,并将这些提示以高维张量的形式提供给SAM,从而增强模型对工业异常的识别能力。而且所提方法不需要对SAM进行任何修改,从而避免复杂的对训练计划的微调需求。在多个数据集上的实验结果表明,所提方法展现出了卓越的性能,与AutoSAM和SAM-EG(SAM with Edge Guidance framework for efficient polyp segmentation)等方法相比,所提方法在mE(mean E-measure)和平均绝对误差(MAE)、Dice和交并比(IoU)上都取得了较好的结果。展开更多
电子轰击型有源像素传感器(Electron Bombarded Active Pixel Sensor,EBAPS)作为一种数字化微光成像器件,以其成像系统具有小型化、低成本和低功耗的优势,以及在昼夜条件下连续拍摄时具备清晰成像的能力,成为微光成像领域的研究热点。...电子轰击型有源像素传感器(Electron Bombarded Active Pixel Sensor,EBAPS)作为一种数字化微光成像器件,以其成像系统具有小型化、低成本和低功耗的优势,以及在昼夜条件下连续拍摄时具备清晰成像的能力,成为微光成像领域的研究热点。文章基于国产某型EBAPS微光器件,利用FPGA作为核心处理器,完成了EBAPS器件驱动电路、图像处理和跟踪电路、显示电路的设计。同时,构建了满足昼夜复用的光学系统,搭建了一种小型化的手持成像跟踪系统。实验结果表明,该EBAPS昼夜成像系统可在1×10^(-4)~1×10^(4) lx照度条件下实现良好的成像和跟踪效果。展开更多
基金The authors gratefully acknowledge the financial support of the National Key Research and Development Program of China(Grant No.2017YFC1600805)the help of Jie Yang in studying convolution neural networks.Trade and manufacturer names are necessary to report factually on the available data。
文摘It is difficult to differentiate small,but harmful,shell fragments of Chinese hickory nuts from their kernels since they are very similar in color.Including shell fragments of Chinese hickory nuts by mistake may create safety hazards for consumers.Therefore,there is a need to develop an effective method to differentiate the shells from the kernels of Chinese hickory nuts.In this study,a deep learning approach based on a two-dimensional convolutional neural network(2D CNN)and long short-term memory(LSTM)integrated with hyperspectral imaging for distinguishing the shells and kernels of Chinese hickory nuts at the pixel level was proposed.Two classical classification methods,principal component analysis-K-nearest neighbors(PCA-KNN)and the support vector machine(SVM),were employed to establish identification models for comparison.The results showed that the 2D CNN-LSTM model achieved the best performance with an overall classification accuracy of 99.0%.Moreover,the shells in mixtures of shells and kernels were detected based on the proposed deep learning method and visualized for subsequent operations for the removal of foreign bodies.
文摘Global cropland monitoring is important when considering tactical strategies for achieving food sustainability. Different global land cover (GLC) datasets providing cropland information have already been published and they are used in many applications. The different data input methods, classification techniques, class definitions and production years among the different GLC datasets make them all independently useful sources of information. This study attempted to produce a cropland agreement level (CAL) analysis based on the integration of several cropland datasets to more accurately estimate cropland area distribution. Estimating cropland area and how it has changed on a national level was done by converting the level of cropland agreement into percentages with an existing cropland fraction map. A pre-analysis showed that the four GLC datasets used in the 2005 and 2010 groups had similar year input data acquisitions. Therefore, we placed these four datasets (GlobCover, MODIS LC, GLCNMO and ESACCI LC) into 2005 and 2010 year-groups and selected them to process dataset integration through a CRISP approach. The results of this process proposed four agreement levels for this CAL analysis, and the model correlation was converted into percentage values. The cropland estimate results from the CAL analysis were observed along with FAO data statistics and showed the highest accuracy, with a 0.70 and 0.71 regression value for 2005 and 2010 respectively. In the cropland area change analysis, this CAL change analysis had the highest level of accuracy when describing the total size of cropland area change from 2005 and 2010 when compared to other individual original GLC datasets.
文摘现有的异常检测方法能在特定应用场景下实现高精度检测,然而这些方法难以适用于其他应用场景,且自动化程度有限。因此,提出一种视觉基础模型(VFM)驱动的像素级图像异常检测方法SSMOD-Net(State Space Model driven-Omni Dimensional Net),旨在实现更精确的工业缺陷检测。与现有方法不同,SSMOD-Net实现SAM(Segment Anything Model)的自动化提示且不需要微调SAM,因此特别适用于需要处理大规模工业视觉数据的场景。SSMOD-Net的核心是一个新颖的提示编码器,该编码器由状态空间模型驱动,能够根据SAM的输入图像动态地生成提示。这一设计允许模型在保持SAM架构不变的同时,通过提示编码器引入额外的指导信息,从而提高检测精度。提示编码器内部集成一个残差多尺度模块,该模块基于状态空间模型构建,能够综合利用多尺度信息和全局信息。这一模块通过迭代搜索,在提示空间中寻找最优的提示,并将这些提示以高维张量的形式提供给SAM,从而增强模型对工业异常的识别能力。而且所提方法不需要对SAM进行任何修改,从而避免复杂的对训练计划的微调需求。在多个数据集上的实验结果表明,所提方法展现出了卓越的性能,与AutoSAM和SAM-EG(SAM with Edge Guidance framework for efficient polyp segmentation)等方法相比,所提方法在mE(mean E-measure)和平均绝对误差(MAE)、Dice和交并比(IoU)上都取得了较好的结果。
