Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Recognizing Potassium Deficiency Symptoms in Soybean with ANN on FPGA

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  Applied Engineering in Agriculture. 38(2): 445-453. (doi: 10.13031/aea.14302) @2022
Authors:   Maicon Aparecido Sartin, Alexandre Cesar Rodrigues da Silva, Claudinei Kappes
Keywords:   Artificial Neural Networks, Digital image processing, Potassium deficiency, Reconfigurable device, Soybean.


Identifying the deficiency of potassium macronutrients in the soybean crop using artificial neural network (ANN).

Image processing based on ANN using a reconfigurable device.

Processing and representation of data in neurons are in floating point.

Abstract. Precision agriculture aims to improve the production of field crops using different techniques to manage the planting stages, such as the monitoring of field crops by images, fertilization control, nutrient analysis of the soil, and pest and weed control. By investigating field images, a plant leaf can be used to identify the lack of nutrients or the presence of diseases. This study developed a system that identifies the macronutrient deficiency of potassium in soybean crops by analyzing leaves. The methodology of this study was developed using different abstractions (Matlab and FPGA) to obtain consolidated results and facilitate low-level abstraction. The main contribution of this study is developing a multilayer artificial neural network system for a reconfigurable device. The developed system was applied in the image segmentation to determine potassium deficiency using soybean leaves and compared with a high-level abstraction system. The results of the reconfigurable device show that the mean hit percentages are 92%, 96%, and 95% in the leaf, trefoil, and field, respectively. The mean square error values were in the range of 10-2 and the quality factor was between 8.5 and 9.0.

(Download PDF)    (Export to EndNotes)