Assessment of Injury by Four Major Pests in Soybean Plants Using Hyperspectral Proximal Imaging

Arthropod pests are among the major problems in soybean production and regular field sampling is required as a basis for decision-making control. However, traditional methods laborious time-consuming. Therefore, our goal to evaluate hyperspectral remote sensing tool establish reflectance patterns from plants infested by various densities of two species stinkbugs (Euschistus heros Diceraeus melacanthus (Hemiptera: Pentatomidae)) caterpillars (Spodoptera eridania Chrysodeixis includens (Lepidoptera: Noctuidae)). Bioassays were carried out greenhouses with potted placed cages 5 0, 2, 5, 10 insects. Plants classified according their reflectance, based on acquisition spectral data before after infestation, using push-broom camera. Infestation did not cause significative differences or non-infested plants. In contrast, caused changes patterns, which deep-learning approach multilayer perceptron artificial neural network. High accuracies achieved when models low (0 + 2) high (5 10) infestation presence absence This study provides an initial assessment apply non-invasive detection method monitor causing economic damage.