Convolutional Neural Networks to Estimate Dry Matter Yield in a Guineagrass Breeding Program Using UAV Remote Sensing.

Forage dry matter is the main source of nutrients in the diet of ruminant animals. Thus, this trait is evaluated in most forage breeding programs with the objective of increasing the yield. Novel solutions combining unmanned aerial vehicles (UAVs) and computer vision are crucial to increase the efficiency of forage breeding programs, to support high-throughput phenotyping (HTP), aiming to estimate parameters correlated to important traits. The main goal of this study was to propose a convolutional neural network (CNN) approach using UAV-RGB imagery to estimate dry matter yield traits in a guineagrass breeding program. For this, an experiment composed of 330 plots of full-sib families and checks conducted at Embrapa Beef Cattle, Brazil, was used. The image dataset was composed of images obtained with an RGB sensor embedded in a Phantom 4 PRO. The traits leaf dry matter yield (LDMY) and total dry matter yield (TDMY) were obtained by conventional agronomic methodology and considered as the ground-truth data. Different CNN architectures were analyzed, such as AlexNet, ResNeXt50, DarkNet53, and two networks proposed recently for related tasks named MaCNN and LF-CNN. Pretrained AlexNet and ResNeXt50 architectures were also studied. Ten-fold cross-validation was used for training and testing the model. Estimates of DMY traits by each CNN architecture were considered as new HTP traits to compare with real traits. Pearson correlation coefficient r between real and HTP traits ranged from 0.62 to 0.79 for LDMY and from 0.60 to 0.76 for TDMY; root square mean error (RSME) ranged from 286.24 to 366.93 kg·ha-1 for LDMY and from 413.07 to 506.56 kg·ha-1 for TDMY. All the CNNs generated heritable HTP traits, except LF-CNN for LDMY and AlexNet for TDMY. Genetic correlations between real and HTP traits were high but varied according to the CNN architecture. HTP trait from ResNeXt50 pretrained achieved the best results for indirect selection regardless of the dry matter trait. This demonstrates that CNNs with remote sensing data are highly promising for HTP for dry matter yield traits in forage breeding programs.

s size with AFM

Atomic Force Microscopy (AFM) can be used to obtain high-resolution topographical images of bacteria revealing surface details and cell integrity. During scanning however, the interactions between the AFM probe and the membrane results in distortion of the images. Such distortions or artifacts are the result of geometrical effects related to bacterial cell height, specimen curvature and the AFM probe geometry. The most common artifact in imaging is surface broadening, what can lead to errors in bacterial sizing. Several methods of correction have been proposed to compensate for these artifacts and in this study we describe a simple geometric model for the interaction between the tip (a pyramidal shaped AFM probe) and the bacterium (Escherichia coli JM-109 strain) to minimize the enlarging effect. Approaches to bacteria immobilization and examples of AFM images analysis are also described.

s size with AFM

Atomic Force Microscopy (AFM) can be used to obtain high-resolution topographical images of bacteria revealing surface details and cell integrity. During scanning however, the interactions between the AFM probe and the membrane results in distortion of the images. Such distortions or artifacts are the result of geometrical effects related to bacterial cell height, specimen curvature and the AFM probe geometry. The most common artifact in imaging is surface broadening, what can lead to errors in bacterial sizing. Several methods of correction have been proposed to compensate for these artifacts and in this study we describe a simple geometric model for the interaction between the tip (a pyramidal shaped AFM probe) and the bacterium (Escherichia coli JM-109 strain) to minimize the enlarging effect. Approaches to bacteria immobilization and examples of AFM images analysis are also described.

Grapevine root distribution in drip and microsprinkler irrigation

Grape (Vitis vinifera L.) yield and its quality are dependent of the status of the root system. Root distribution information is also valuable for soil and water management. An analysis of methods to evaluate the root distribution of grapevines for both, drip and microsprinkler irrigation in a Typic Acrustox is presented for the table grape cv. Italia grafted on the rootstock IAC-313, in northeastern Brazil. Measured root parameters using the monolith method were root dry weight (Dw) and root length density (Lv), while root area (Ap) was estimated using the soil profile method in combination with digital image analysis. For both irrigation systems, roots were present to the 1 m soil depth and extended laterally to 1 m distance from the trunk, but grapevines irrigated by microsprinkler showed greater root presence as the distance from the trunk increased. Values of Ap were reasonably well correlated to Dw and Lv. However, correlation values were higher when fractional root distribution was used. The soil profile method in combination with image analysis techniques, allows proper grapevine root distribution evaluation.

A produção de uva (Vitis vinifera L.) em termos quantitativos e qualitativos depende do estado das raízes. Além disso, informações sobre a distribuição radicular são úteis para o manejo de solo e água. Por isso, uma análise de métodos para a avaliação da distribuição radicular de videiras cv. Itália / IAC 313 num Latossolo Vermelho Amarelo irrigadas por gotejamento e microaspersão foi realizada em Petrolina PE e Juazeiro - BA, no Vale do São Francisco. Os parâmetros medidos pelo método do monolito foram a matéria seca (Dw) e densidade de comprimento de raízes (Lv), enquanto a área de raízes (Ap) foi estimada pelo método do perfil de solo combinado com a análise de imagens digitais. Para ambos os sistemas de irrigação, as raízes estiveram presentes até 1 m de profundidade e estenderam-se lateralmente até 1 m de distância do tronco, mas as videiras irrigadas por microaspersão apresentaram uma maior presença de raízes com o aumento da distância do tronco. Os valores de Ap apresentaram uma boa correlação com Dw e Lv, mas essa correlação foi maior quando se utilizou a distribuição fracional de cada parâmetro. O método do perfil auxiliado pela análise de imagem digital permite a avaliação da distribuição radicular.

Grapevine root distribution in drip and microsprinkler irrigation

Grape (Vitis vinifera L.) yield and its quality are dependent of the status of the root system. Root distribution information is also valuable for soil and water management. An analysis of methods to evaluate the root distribution of grapevines for both, drip and microsprinkler irrigation in a Typic Acrustox is presented for the table grape cv. Italia grafted on the rootstock IAC-313, in northeastern Brazil. Measured root parameters using the monolith method were root dry weight (Dw) and root length density (Lv), while root area (Ap) was estimated using the soil profile method in combination with digital image analysis. For both irrigation systems, roots were present to the 1 m soil depth and extended laterally to 1 m distance from the trunk, but grapevines irrigated by microsprinkler showed greater root presence as the distance from the trunk increased. Values of Ap were reasonably well correlated to Dw and Lv. However, correlation values were higher when fractional root distribution was used. The soil profile method in combination with image analysis techniques, allows proper grapevine root distribution evaluation.

A produção de uva (Vitis vinifera L.) em termos quantitativos e qualitativos depende do estado das raízes. Além disso, informações sobre a distribuição radicular são úteis para o manejo de solo e água. Por isso, uma análise de métodos para a avaliação da distribuição radicular de videiras cv. Itália / IAC 313 num Latossolo Vermelho Amarelo irrigadas por gotejamento e microaspersão foi realizada em Petrolina PE e Juazeiro - BA, no Vale do São Francisco. Os parâmetros medidos pelo método do monolito foram a matéria seca (Dw) e densidade de comprimento de raízes (Lv), enquanto a área de raízes (Ap) foi estimada pelo método do perfil de solo combinado com a análise de imagens digitais. Para ambos os sistemas de irrigação, as raízes estiveram presentes até 1 m de profundidade e estenderam-se lateralmente até 1 m de distância do tronco, mas as videiras irrigadas por microaspersão apresentaram uma maior presença de raízes com o aumento da distância do tronco. Os valores de Ap apresentaram uma boa correlação com Dw e Lv, mas essa correlação foi maior quando se utilizou a distribuição fracional de cada parâmetro. O método do perfil auxiliado pela análise de imagem digital permite a avaliação da distribuição radicular.

Distribuição do sistema radicular de uma cultura de aveia forrageira

Forage oat root distribution was evaluated through the application of the methods of the volumetric ring, of the nail plate, of the auger, and of image processing. The crop was grown on a Dark Red Latosol, submitted to two different treatments in relation to water avalilability: 1. Mantaining at least 50% of the soil available water, and 2. Mantaining 80%. Three samplings were perfoemed, 25, 60, and 115 days after emergence. Oat roots were concentrated in the top soil layer, about 90% from 0 to 20 cm depth. Results of root quntification show, in general, distinct with respect to treatment, methodology and sampling date. Correlations were found between data of dry root weigth and root presence. The root variability in the soil depended on the method of evaluation. The lowest coefficients of variation were found for the auger method, indicating that the lowest variability of root distribution has a greater relation to sample volume then on the number of sampling points.

Avaliou-se a distribuição das raízes de uma aveia forrageira através da aplicação dos métodos do anel volumétrico, da placa com pregos, do trado e do auxiliado por processamento de imagens. A aveia forrageira foi cultivada em latossolo vermelho-escuro, sob dois tratamentos distintos em função da manutenção da capacidade de água disponível no solo (CAD): tratamento 1 - 50% da CAD; tratamento 2 - 80% da CAD. Foram realizadas 3 amostragens, aos 25, 60 e 115 dias após a emergência. As raízes da aveia forrageira concentraram-se nas camadas iniciais do solo, cerca de 90% entre 0 a 20 cm. Os resultados encontrados para a quantificação das raízes variaram com o tratamento, a metodologia e a época de amostragem. Verificou-se a existência de correlações entre os resultados determinados a partir das grandezas massa seca de raiz e presença radicular. A variabilidade da distribuição das raízes no solo foi dependente do método de avaliação utilizado. Os menores valores dos coeficientes de variação foram observados para o método do trado, indicando que a menor variabilidade da distribuição das raízes tem maior relação com o volume da amostra do que com o número de pontos amostrados.

Distribuição do sistema radicular de uma cultura de aveia forrageira

Forage oat root distribution was evaluated through the application of the methods of the volumetric ring, of the nail plate, of the auger, and of image processing. The crop was grown on a Dark Red Latosol, submitted to two different treatments in relation to water avalilability: 1. Mantaining at least 50% of the soil available water, and 2. Mantaining 80%. Three samplings were perfoemed, 25, 60, and 115 days after emergence. Oat roots were concentrated in the top soil layer, about 90% from 0 to 20 cm depth. Results of root quntification show, in general, distinct with respect to treatment, methodology and sampling date. Correlations were found between data of dry root weigth and root presence. The root variability in the soil depended on the method of evaluation. The lowest coefficients of variation were found for the auger method, indicating that the lowest variability of root distribution has a greater relation to sample volume then on the number of sampling points.

Avaliou-se a distribuição das raízes de uma aveia forrageira através da aplicação dos métodos do anel volumétrico, da placa com pregos, do trado e do auxiliado por processamento de imagens. A aveia forrageira foi cultivada em latossolo vermelho-escuro, sob dois tratamentos distintos em função da manutenção da capacidade de água disponível no solo (CAD): tratamento 1 - 50% da CAD; tratamento 2 - 80% da CAD. Foram realizadas 3 amostragens, aos 25, 60 e 115 dias após a emergência. As raízes da aveia forrageira concentraram-se nas camadas iniciais do solo, cerca de 90% entre 0 a 20 cm. Os resultados encontrados para a quantificação das raízes variaram com o tratamento, a metodologia e a época de amostragem. Verificou-se a existência de correlações entre os resultados determinados a partir das grandezas massa seca de raiz e presença radicular. A variabilidade da distribuição das raízes no solo foi dependente do método de avaliação utilizado. Os menores valores dos coeficientes de variação foram observados para o método do trado, indicando que a menor variabilidade da distribuição das raízes tem maior relação com o volume da amostra do que com o número de pontos amostrados.