Estimation of peanut yield using unmanned aircraft systems and machine learning

Authors: Pugh, Nicholas - Ace; Young, Andrew; OJHA, MANISHA - New Mexico State University; Sanchez, Jacobo; Emendack, Yves; Xin, Zhanguo; PUPPALA, NAVEEN - New Mexico State University

Submitted to: ASA-CSSA-SSSA Annual Meeting Abstracts
Publication Type: Abstract Only
Publication Acceptance Date: 6/19/2023
Publication Date: 10/29/2023
Citation: Pugh, N.A., Young, A.W., Ojha, M., Sanchez, J., Emendack, Y., Xin, Z., Puppala, N. 2023. Estimation of peanut yield using unmanned aircraft systems and machine learning. ASA-CSSA-SSSA Annual Meeting Abstracts.

Interpretive Summary:

Technical Abstract: Peanuts are an important crop globally, both nutritionally and economically. The largest bottleneck in peanut improvement pipelines is phenotyping, but remote and proximal sensing technologies present a possible method to improve the efficiency of peanut breeding programs. However, the subterranean growth of the legumes can make it difficult or impossible to make predictions about peanut yield directly. Thus, the objectives of this study were: i) to create high-resolution multitemporal growth curves for vegetation indices, canopy volume, and other critical traits extracted from imagery collected via unmanned aerial vehicles in peanuts; ii) to derive latent phenotypes from the growth curves and other data; and iii) to use machine learning methodologies to create linear regression models that allow researchers to predict yield. Currently, ground cover percentage shows a 0.77 (r) correlation at 71 days after planting (DAP) with final yield, and excess greenness index (ExG) has a 0.57 (r) correlation with final yield at 78 DAP.