Digital count of Sunflower plants at emergence from very low altitude using UAV images

Abstract

The implementation of new technologies for agriculture has gained relevance in the agricultural sector. In this sense, during the last decade, the use of aerial images has allowed to study the different factors related to production with a high level of detail, unlike traditional monitoring techniques, since these allow large areas to be analyzed in a short time. Aerial images are commonly obtained from satellite platforms because they have a low cost, but low spatial and temporal resolution limit their use in agricultural applications. For this reason, the use of unmanned aerial vehicles (UAV) has played a leading role in the study of agronomic variables of interest, because it allows obtaining information at any time and at a higher resolution. However, these images can offer more information that must be examined through a set of analysis techniques. An application that has been little explored corresponds to the population count, a factor that is determinant to obtain production estimates, but the difficulty to properly segment the rest of the information in the images has posed a challenge. For this reason, the following work presents a methodology based on spectral indices and digital image analysis to perform population counts in sunflower plants. Results indicate that it is possible to estimate the number of plants in the image with an error of 10%.