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.

Cotton flower detection using aerial color images

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

Citation:  2017 ASABE Annual International Meeting  1701080.(doi:10.13031/aim.201701080)
Authors:   Rui Xu, Andrew Paterson, Changying Li
Keywords:   Click here to enter keywords and key phrases, separated by commas, with a period at the end

Abstract. Cotton flower is important to cotton yield because pollinated flowers form cotton bolls. Monitoring flower development can provide useful information for production management and estimation of the yield. A data processing pipeline was developed to detect and count cotton flowers using color images taken by an unmanned aerial vehicle. A convolutional neural network (CNN) was built to classify potential flower images that were selected with a thresholding method based on the pixel intensity. The trained CNN had a 4.5% false negative rate and 5.1% false positive rate for the training images. The flower counts from images were significantly underestimated compared to the manual counts due to the inability to detect the flowers that are blocked by leaves. However, the overall trend of the image counting is consistent with the manual counting. The seasonal bloom count was correlated with the cotton yield, which means seasonal bloom count can be used as an indicator to estimate yield.

Keywords. Cotton, convolutional neural network, flower, unmanned aerial vehicle, phenotyping

(Download PDF)    (Export to EndNotes)