Using multispectral sensors attached to unmanned aerial vehicles (UAVs) can assist in the collection of morphological and physiological information from several crops. This approach, also known as high-throughput phenotyping, combined with data processing by machine learning (ML) algorithms, can provide fast, accurate, and large-scale discrimination of genotypes in the field, which is crucial for improving the efficiency of breeding programs. Despite their importance, studies aimed at accurately classifying sorghum hybrids using spectral variables as input sets in ML models are still scarce in the literature. Against this backdrop, this study aimed: (I) to discriminate sorghum hybrids based on canopy reflectance in different spectral bands (SB) and vegetation indices (VIs); (II) to evaluate the performance of ML algorithms in classifying sorghum hybrids; (III) to evaluate the best dataset input for the algorithms. A field experiment was carried out in the 2022 crop season in a randomized block design with three replications and six sorghum hybrids. At 60 days after crop emergence, a flight was carried out over the experimental area using the Sensefly eBee real time kinematic. The spectral bands (SB) acquired by the sensor were: blue (475 nm, B_475), green (550 nm, G_550), red (660 nm, R_660), Rededge (735 nm, RE_735) e NIR (790 nm, NIR_790). From the SB acquired, vegetation indices (VIs) were calculated. Data were submitted to ML classification analysis, in which three input settings (using only SB, using only VIs, and using SB + VIs) and six algorithms were tested: artificial neural networks (ANN), support vector machine (SVM), J48 decision trees (J48), random forest (RF), REPTree (DT) and logistic regression (LR, conventional technique used as a control). There were differences in the spectral signature of each sorghum hybrid, which made it possible to differentiate them using SBs and VIs. The ANN algorithm performed best for the three accuracy metrics tested, regardless of the input used. In this case, the use of SB is feasible due to the speed and practicality of analyzing the data, as it does not require calculations to perform the VIs. RF showed better accuracy when VIs were used as an input. The use of VIs provided the best performance for all the algorithms, as did the use of SB + VIs which provided good performance for all the algorithms except RF. Using ML algorithms provides accurate identification of the hybrids, in which ANNs using only SB and RF using VIs as inputs stand out (above 55 for CC, above 0.4 for kappa and around 0.6 for F-score). There were differences in the spectral signature of each sorghum hybrid, which makes it possible to differentiate them using wavelengths and vegetation indices. Processing the multispectral data using machine learning techniques made it possible to accurately differentiate the hybrids, with emphasis on artificial neural networks using spectral bands as inputs and random forest using vegetation indices as inputs.

The integration multispectral sensors with machine learning algorithms has demonstrated increasing efficacy in the classification of various maize morphophysiological characteristics. The hypothesis of this study is that maize plants subjected to different irrigation management practices exhibit distinct spectral behaviors, allowing for their classification through machine learning modeling. Thus, the objective of this study is to classify maize hybrids in different irrigation management practices using multispectral images. This involves identifying the most effective machine learning algorithms and inputs variables that enhance model performance for accurate classification. The experiment was conducted at the experimental facility of the Federal University of Mato Grosso do Sul, in Chapadão do Sul – MS. Seven hybrids were evaluated: H1 (AS 1868), H2 (DKB 360), H3 (FS 615 PWU), H4 (K 7510 VIP3), H5 (NK 520 VIP3), H6 (P 3858 PWU), and H7 (SS 182E VIP3). These hybrids were subjected to irrigation and non-irrigation management practices. Sixty days after crop emergence, images were captured in the blue (475 nm, B_475), green (550 nm, G_550), red (660 nm, R_660), red edge (735 nm, RE_735), and near-infrared (790 nm, NIR_790) bands using the Sensefly eBee RTK fixed-wing Remotely Piloted Aircraft, equipped with a Parrot Sequoia multispectral sensor and RTK (Real-Time Kinematics) technology. Through the collected band data, the ESRI ArcGIS 10.5 geographic information system software was used to calculate 41 vegetation indices (VIs). Data were analyzed using machine learning techniques, testing six algorithms: Logistic Regression (RL), REPTree (DT), J48 Decision Trees (J48), Random Forest (RF), Artificial Neural Networks (ANN) and Support Vector Machine (SVM). Three accuracy metrics were utilized to evaluate the algorithms in the classification of irrigation management: correct classifications (CC), Kappa coefficient and F-Score. The ANN and RF algorithms demonstrated better accuracy in classifying maize hybrids with respect to irrigation management. The use of Vegetation Indices (IVs) and Spectral Bands + Vegetation Indices (SB+IVs) enhanced performance of these algorithms.

Maize cultivation in tropical Oxisols during the second growing season faces significant climatic risks, where spatial heterogeneity in soil water retention often dictates economic viability. This study integrated a trimodal sensing approach, combining multispectral, thermal, and LiDAR data, with proximal physiological measurements to map isohydric responses and yield variability. Conducted in the Brazilian Cerrado, the research monitored a one-hectare maize field using UAV-based sensors alongside ground truth evaluations of gas exchange, leaf water potential, and soil moisture. Results revealed high yield variability (6.6 to 13.4 Mg ha⁻¹) primarily governed by clay content-mediated water availability. Maize exhibited strict isohydric behavior, maintaining homeostatic leaf water potential through preventive stomatal closure, which limited CO₂ assimilation in zones with lower water retention. A significant statistical decoupling was observed between plant height and final grain yield, as water stress impacted reproductive stages more severely than vegetative growth. Furthermore, the Temperature Vegetation Dryness Index (TVDI) served as a robust proxy for biomass vigor rather than mere water deficit. These results confirm that yield variability in tropical Oxisols was not a product of hydraulic failure, but rather a consequence of carbon limitation necessitated by the crop’s conservative hydraulic management to maintain leaf water potential within safe thresholds.

Novel methodologies for phenotypic evaluation in maize have been developed through the integration of advanced sensing technologies and machine learning algorithms. The aim of this study was to identify the most accurate machine learning algorithm for the classification of maize hybrids and to determine the optimal input data to enhance model performance. Seven maize hybrids were used in the experiment. After 60 d of crop emergence, the remotely piloted aircraft SenseFly® eBee RTK was used to obtain reflectance values at the following spectral bands (SB): blue (475 nm, B_475), green (550 nm, G_550), red (660 nm, R_660), red edge (735 nm, RE_735) and near-infrared (790 nm, NIR_790). Following the acquisition of spectral band (SB) data, vegetation indices (VIs) were calculated. The resulting dataset was subsequently analyzed using machine learning techniques, evaluating six algorithms: artificial neural networks (ANN), J48 decision trees (J48), REPTree (DT), random forest (RF), support vector machine (SVM) and logistic regression (LR) as the baseline model. Three accuracy metrics were used to evaluate the performance of the algorithms in classifying maize hybrids: correct classifications (CC), Kappa coefficient, and F-score. Among the algorithms tested, ANN showed the highest performance in all three metrics, proving its superiority and potential for real-world applications. Although all three input configurations enhanced classification accuracy for ANN algorithm, the optimal approach is to use only SB as input due to reduced data processing time and increased simplicity.