A fast and robust method for plant quantity estimation in sunflower and maize at different seedling stages using high-resolution UAV RGB imagery

Keywords:  UAV, Plant quantity estimation, Peak detecting algorithm
Session Type: Virtual Paper
Day: Wednesday
Session Start / End Time: 4/7/2021 08:00 AM (Pacific Time (US & Canada)) - 4/7/2021 10:20 AM (Pacific Time (US & Canada))
Room: Virtual 29

Abstract

The acquisition of crop plant quantity is key to improving field decision-making at the seedling stage. Therefore, it is necessary to count the number of plants under different climate and soil conditions for effective management and high yield guarantee. In recent years, the development of high-resolution sensors has provided new opportunities to estimate the number of plants. However, plant quantity estimation using remote sensing still faces great challenges.
Previous studies using machine learning methods only focused on the plant quantity of single crop. When the target crop changes, the model needs to be re-trained, which requires extensive training data considerable time and space. Therefore, a universal method is needed to estimate the plant quantity in the field. Furthermore, Liu et al. (2018) only studied the plant count in the early stage of the crop seedling (1-3 leaves). In the late seedling stage (more than 3 leaves), the loss of precision caused by the overlapping between plants is still a difficult problem in estimating the number of plants.
In this study, our primary objective is to overcome the plant overlapping problem and automatically count row crop plants. Specifical challenges to be tackled are: (1) to develop a fast and generic machine vision method to estimate equidistant crop plant quantity based on high-resolution RGB images acquired by unmanned aerial vehicle (UAV); (2) to test this method on field crops with different numbers of leaves at the seedling stage; (3) to investigate the influence of image resolution on crop quantity estimation