Oyinlola, Adeoluwa; Odedeyi, Temitope; (2025) Can ML Enhance the Accuracy of Crop Quality Assessment With Radio Frequency Reflectometry? IEEE Transactions on AgriFood Electronics 10.1109/tafe.2025.3624544. (In press).

Preview

Text TAFE3624544-Peer Reviewed Accepted Manuscript.pdf - Accepted Version Download (6MB) | Preview

Abstract

This article presents a novel approach that integrates AI/ML algorithms-linear regression, random forest, and principal component analysis (PCA) to enhance the predictive accuracy of radio frequency reflectometric (RF-R) analysis for crop quality estimation. Using 131 samples of cassava roots, RF-R measurements were carried out within 30 kHz–200 MHz frequency range to estimate dry matter as a proxy for starch content. Multiple modeling strategies were evaluated, including iterative linear regression, random forest, and PCA, which were applied to both single-frequency and multifrequency combinations. A single-frequency analysis from simple linear regression achieved an R2 of 0.64, while, for the first time, we show that combining two and three frequencies for a random forest model enhances prediction accuracy by 23.4%, highlighting the potential for improved crop quality assessment. Overall, this work demonstrates that combining RF-R with ML improves the specificity and robustness of quality prediction, establishing a foundation for application in real-time, nondestructive crop assessment.

Download activity - last month

Export as CSVJSONXML

Download activity - last 12 months

Export as CSVJSONXML

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item