AUTHOR=Bai Dong , Li Delin , Zhao Chaosen , Wang Zixu , Shao Mingchao , Guo Bingfu , Liu Yadong , Wang Qi , Li Jindong , Guo Shiyu , Wang Ruizhen , Li Ying-hui , Qiu Li-juan , Jin Xiuliang 

TITLE=Estimation of soybean yield parameters under lodging conditions using RGB information from unmanned aerial vehicles

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 13 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2022.1012293

DOI=10.3389/fpls.2022.1012293

ISSN=1664-462X

ABSTRACT=The estimation of yield parameters based on early data is helpful for agricultural policymakers and food security. The development of unmanned aerial vehicle (UAV) platforms and sensor technology help to collect data at a lower cost with a higher spatial and temporal resolution. However, previous studies did not fully explore how to estimate crop yield parameters, particularly lodging conditions, using different information from UAV red green blue (RGB) images. Therefore, the objective of this study was to estimate the yield parameters of soybean (Glycine max (L.) Merr.)  under lodging conditions using different information, including vegetation index, canopy cover, digital elevation model, and texture indicators, from UAV RGB images and to assess the effect of different conditions of lodging for yield parameters. In this study, 17 time point data throughout the soybean growing season in Nanchang, Jiangxi Province, China, were collected, and the vegetation index, texture information, canopy cover, and crop height were obtained by image processing. After that, partial least squares regression (PLSR), logistic regression (Logistic), random forest regression (RFR), support vector machine regression (SVM), and deep learning neural network (DNN) were used to estimate the yield parameters. The results can be summarized as follows: (1) Lodging had a high effect on both the grain number of seeds per plant and the grain weight per plant. (2) The most suitable time point to estimate the yield was 48 days after sowing, which was when most of the soybean cultivars flowered. (3) The multiple data fusion improved the accuracy of estimating the yield parameters, and the texture information has a high potential to contribute to the estimation of yields, and (4) The DNN model showed the best accuracy of training (R2=0.66 rRMSE=32.62%) and validation (R2=0.50, rRMSE=43.71%) datasets. These results provide insights into both best estimate period selection and early yield estimation under lodging condition when using remote sensing.