为在田间复杂环境中实现对杂草和玉米植株准确实例分割和叶龄识别获取,提出一种基于改进掩码区域卷积神经网络(Mask Regions with convolutional neural network features,Mask R-CNN)的植物叶龄获取方法。具体实施为构建包含不同天气(...为在田间复杂环境中实现对杂草和玉米植株准确实例分割和叶龄识别获取,提出一种基于改进掩码区域卷积神经网络(Mask Regions with convolutional neural network features,Mask R-CNN)的植物叶龄获取方法。具体实施为构建包含不同天气(晴天、阴天、雨后)和不同采集角度(俯视、30°斜视、45°斜视)数据集,增强数据并用作网络输入。通过更换3个特征提取网络(ResNet-50、ResNet-101、MobileNetv2)、搭建多种不同尺寸区域建议框、非极大值抑制法(Non-maximum suppression,NMS)更换为Soft-NMS算法、RoIAlign代替RoI Pooling方法提高模型精度。测试田间复杂环境下杂草和玉米图像。结果表明,以ResNet-101为特征提取网络的改进深度学习模型具有良好分割性能和鲁棒性,阴天检测精度高于晴天和雨后,30°斜视检测效果优于45°斜视和俯视。分割模型AP50为0.730,高于现有DeepMask、MNC、Mask R-CNN分割模型精度,表明该方法可提高对杂草和玉米植株的实例分割和叶龄识别精度。展开更多
Harnessing natural variation in photosynthetic capacity is a promising route toward yield increases, but physiological phenotyping is still too laborious for large-scale genetic screens. Here, we evaluate the potentia...Harnessing natural variation in photosynthetic capacity is a promising route toward yield increases, but physiological phenotyping is still too laborious for large-scale genetic screens. Here, we evaluate the potential of leaf reflectance spectroscopy to predict parameters of photosynthetic capacity in Brassica oleracea and Zea mays, a C3 and a C4 crop, respectively. To this end, we systematically evaluated properties of reflectance spectra and found that they are surprisingly similar over a wide range of species. We assessed the performance of a wide range of machine learning methods and selected recursive feature elimination on untransformed spectra followed by partial least squares regression as the preferred algorithm that yielded the highest predictive power. Learning curves of this algorithm suggest optimal species-specific sample sizes. Using the Brassica relative Moricandia, we evaluated the model transferability between spe- cies and found that cross-species performance cannot be predicted from phylogenetic proximity. The final intra-species models predict crop photosynthetic capacity with high accuracy. Based on the estimated model accuracy, we simulated the use of the models in selective breeding experiments, and showed that high-throughput photosynthetic phenotyping using our method has the potential to greatly improve breeding success. Our results indicate that leaf reflectance phenotyping is an efficient method for improving crop photosynthetic capacity.展开更多
Assessment of vegetation biochemical and biophysical variables is useful when developing indicators for biodiversity monitoring and climate change studies.Here,we compared a radiative transfer model(RTM)inversion by m...Assessment of vegetation biochemical and biophysical variables is useful when developing indicators for biodiversity monitoring and climate change studies.Here,we compared a radiative transfer model(RTM)inversion by merit function and five machine learning algorithms trained on an RTM simulated dataset predicting the three plant traits leaf chlorophyll content(LCC),canopy chlorophyll content(CCC),and leaf area index(LAI),in a mixed temperate forest.The accuracy of the retrieval methods in predicting these three plant traits with spectral data from Sentinel-2 acquired on 13 July 2017 over Bavarian Forest National Park,Germany,was evaluated using in situ measurements collected contemporaneously.The RTM inversion using merit function resulted in estimations of LCC(R^(2)=0.26,RMSE=3.9µg/cm^(2)),CCC(R^(2)=0.65,RMSE=0.33 g/m^(2)),and LAI(R^(2)=0.47,RMSE=0.73 m^(2)/m^(2)),comparable to the estimations based on the machine learning method Random forest regression of LCC(R^(2)=0.34,RMSE=4.06µg/cm^(2)),CCC(R^(2)=0.65,RMSE=0.34 g/m^(2)),and LAI(R^(2)=0.47,RMSE=0.75 m^(2)/m^(2)).Several of the machine learning algorithms also yielded accuracies and robustness similar to the RTM inversion using merit function.The performance of regression methods trained on synthetic datasets showed promise for fast and accurate mapping of plant traits accross different plant functional types from remote sensing data.展开更多
Acquisition of plant phenotypic information facilitates plant breeding, sheds light on gene action, and can be applied to optimize the quality of agricultural and forestry products. Because leaves often show the faste...Acquisition of plant phenotypic information facilitates plant breeding, sheds light on gene action, and can be applied to optimize the quality of agricultural and forestry products. Because leaves often show the fastest responses to external environmental stimuli, leaf phenotypic traits are indicators of plant growth,health, and stress levels. Combination of new imaging sensors, image processing, and data analytics permits measurement over the full life span of plants at high temporal resolution and at several organizational levels from organs to individual plants to field populations of plants. We review the optical sensors and associated data analytics used for measuring morphological, physiological, and biochemical traits of plant leaves on multiple scales. We summarize the characteristics, advantages and limitations of optical sensing and data-processing methods applied in various plant phenotyping scenarios. Finally, we discuss the future prospects of plant leaf phenotyping research. This review aims to help researchers choose appropriate optical sensors and data processing methods to acquire plant leaf phenotypes rapidly,accurately, and cost-effectively.展开更多
Plant recognition has great potential in forestry research and management.A new method combined back propagation neural network and radial basis function neural network to identify tree species using a few features an...Plant recognition has great potential in forestry research and management.A new method combined back propagation neural network and radial basis function neural network to identify tree species using a few features and samples.The process was carried out in three steps:image pretreatment,feature extraction,and leaf recognition.In the image pretreatment processing,an image segmentation method based on hue,saturation and value color space and connected component labeling was presented,which can obtain the complete leaf image without veins and back-ground.The BP-RBF hybrid neural network was used to test the influence of shape and texture on species recogni-tion.The recognition accuracy of different classifiers was used to compare classification performance.The accuracy of the BP-RBF hybrid neural network using nine dimensional features was 96.2%,highest among all the classifiers.展开更多
In the agricultural industry,rice infections have resulted in significant productivity and economic losses.The infections must be recognized early on to regulate and mitigate the effects of the attacks.Early diagnosis...In the agricultural industry,rice infections have resulted in significant productivity and economic losses.The infections must be recognized early on to regulate and mitigate the effects of the attacks.Early diagnosis of disease severity effects or incidence can preserve production from quantitative and qualitative losses,reduce pesticide use,and boost ta country’s economy.Assessing the health of a rice plant through its leaves is usually done as a manual ocular exercise.In this manuscript,three rice plant diseases:Bacterial leaf blight,Brown spot,and Leaf smut,were identified using the Alexnet Model.Our research shows that any reduction in rice plants will have a significant beneficial impact on alleviating global food hunger by increasing supply,lowering prices,and reducing production's environmental impact that affects the economy of any country.Farmers would be able to get more exact and faster results with this technology,allowing them to administer the most acceptable treatment available.By Using Alex Net,the proposed approach achieved a 99.0%accuracy rate for diagnosing rice leaves disease.展开更多
文摘Harnessing natural variation in photosynthetic capacity is a promising route toward yield increases, but physiological phenotyping is still too laborious for large-scale genetic screens. Here, we evaluate the potential of leaf reflectance spectroscopy to predict parameters of photosynthetic capacity in Brassica oleracea and Zea mays, a C3 and a C4 crop, respectively. To this end, we systematically evaluated properties of reflectance spectra and found that they are surprisingly similar over a wide range of species. We assessed the performance of a wide range of machine learning methods and selected recursive feature elimination on untransformed spectra followed by partial least squares regression as the preferred algorithm that yielded the highest predictive power. Learning curves of this algorithm suggest optimal species-specific sample sizes. Using the Brassica relative Moricandia, we evaluated the model transferability between spe- cies and found that cross-species performance cannot be predicted from phylogenetic proximity. The final intra-species models predict crop photosynthetic capacity with high accuracy. Based on the estimated model accuracy, we simulated the use of the models in selective breeding experiments, and showed that high-throughput photosynthetic phenotyping using our method has the potential to greatly improve breeding success. Our results indicate that leaf reflectance phenotyping is an efficient method for improving crop photosynthetic capacity.
文摘Assessment of vegetation biochemical and biophysical variables is useful when developing indicators for biodiversity monitoring and climate change studies.Here,we compared a radiative transfer model(RTM)inversion by merit function and five machine learning algorithms trained on an RTM simulated dataset predicting the three plant traits leaf chlorophyll content(LCC),canopy chlorophyll content(CCC),and leaf area index(LAI),in a mixed temperate forest.The accuracy of the retrieval methods in predicting these three plant traits with spectral data from Sentinel-2 acquired on 13 July 2017 over Bavarian Forest National Park,Germany,was evaluated using in situ measurements collected contemporaneously.The RTM inversion using merit function resulted in estimations of LCC(R^(2)=0.26,RMSE=3.9µg/cm^(2)),CCC(R^(2)=0.65,RMSE=0.33 g/m^(2)),and LAI(R^(2)=0.47,RMSE=0.73 m^(2)/m^(2)),comparable to the estimations based on the machine learning method Random forest regression of LCC(R^(2)=0.34,RMSE=4.06µg/cm^(2)),CCC(R^(2)=0.65,RMSE=0.34 g/m^(2)),and LAI(R^(2)=0.47,RMSE=0.75 m^(2)/m^(2)).Several of the machine learning algorithms also yielded accuracies and robustness similar to the RTM inversion using merit function.The performance of regression methods trained on synthetic datasets showed promise for fast and accurate mapping of plant traits accross different plant functional types from remote sensing data.
基金supported by the National Natural Science Foundation of China (32171790 and 32171818)Jiangsu Province Modern Agricultural Machinery Equipment and Technology Demonstration Promotion Project (NJ2020-18)+2 种基金Key Research and Development Program of Jiangsu Province (BE2021307)Qinglan Project Foundation of Jiangsu province333 Project of Jiangsu Province。
文摘Acquisition of plant phenotypic information facilitates plant breeding, sheds light on gene action, and can be applied to optimize the quality of agricultural and forestry products. Because leaves often show the fastest responses to external environmental stimuli, leaf phenotypic traits are indicators of plant growth,health, and stress levels. Combination of new imaging sensors, image processing, and data analytics permits measurement over the full life span of plants at high temporal resolution and at several organizational levels from organs to individual plants to field populations of plants. We review the optical sensors and associated data analytics used for measuring morphological, physiological, and biochemical traits of plant leaves on multiple scales. We summarize the characteristics, advantages and limitations of optical sensing and data-processing methods applied in various plant phenotyping scenarios. Finally, we discuss the future prospects of plant leaf phenotyping research. This review aims to help researchers choose appropriate optical sensors and data processing methods to acquire plant leaf phenotypes rapidly,accurately, and cost-effectively.
基金This work is supported by the Fundamental Research Funds for the Central Universities(No.2572020BC07)the Project of National Science Foundation of China(No.31570712).
文摘Plant recognition has great potential in forestry research and management.A new method combined back propagation neural network and radial basis function neural network to identify tree species using a few features and samples.The process was carried out in three steps:image pretreatment,feature extraction,and leaf recognition.In the image pretreatment processing,an image segmentation method based on hue,saturation and value color space and connected component labeling was presented,which can obtain the complete leaf image without veins and back-ground.The BP-RBF hybrid neural network was used to test the influence of shape and texture on species recogni-tion.The recognition accuracy of different classifiers was used to compare classification performance.The accuracy of the BP-RBF hybrid neural network using nine dimensional features was 96.2%,highest among all the classifiers.
文摘In the agricultural industry,rice infections have resulted in significant productivity and economic losses.The infections must be recognized early on to regulate and mitigate the effects of the attacks.Early diagnosis of disease severity effects or incidence can preserve production from quantitative and qualitative losses,reduce pesticide use,and boost ta country’s economy.Assessing the health of a rice plant through its leaves is usually done as a manual ocular exercise.In this manuscript,three rice plant diseases:Bacterial leaf blight,Brown spot,and Leaf smut,were identified using the Alexnet Model.Our research shows that any reduction in rice plants will have a significant beneficial impact on alleviating global food hunger by increasing supply,lowering prices,and reducing production's environmental impact that affects the economy of any country.Farmers would be able to get more exact and faster results with this technology,allowing them to administer the most acceptable treatment available.By Using Alex Net,the proposed approach achieved a 99.0%accuracy rate for diagnosing rice leaves disease.
