Published January 5, 2021 | Version v1
Conference paper Open

Cotton Stand Count Using UAS Imagery and Deep Learning

Creators

  • 1. Department of Plant & Soil Science, Texas Tech University Lubbock, TX, USA
  • 2. Department of Natural Resources and Agricultural Engineering, Faculty of Agriculture, Damanhour University, Damanhour, Egypt

Description

Stand count is critical for growers to make decisions for replanting and other site-specific management to avoid yield
loss. This study applied and compared two object detection models, MobileNets and CenterNet, in cotton stand count
using unmanned aerial system (UAS) images. The results showed that the overall mean precision and recall for the
CenterNet model were higher than those for the MobileNets model. The CenterNet trained model had an R2 value of
0.60 and the MobileNets trained model had an R2 value of 0.48. The accuracy for the CenterNet model was 61% and
for the MobileNets model was 46%. The results indicate that the CenterNet model has a better overall performance
on cotton plant detection and counting.
 

Notes

Cotton, Stand count, UAS, MobileNets, CenterNet, Deep Learning

Files

Cotton Stand Count Using UAS Imagery and Deep Learning.pdf

Files (667.9 kB)