Journal of Experimental Agriculture International

In maize crops, the fall armyworm (FAW) (Spodoptera frugiperda) is a major invasive pest that causes significant economic losses. Traditional visual detection of pest infestations is labor-intensive, but spectral data related to plant damage can aid in automated detection. The main purpose of the study is to identify FAW infestation levels in maize using hyperspectral proximal remote sensing and machine learning. Nylon net cages were placed over maize plants in a field, and artificial FAW infestations were created. A handheld Apogee multispectral spectroradiometer (350–1150 nm) measured reflectance changes at varying FAW densities with a 1 nm spectral resolution.  Reflectance result revealed that the visible zone (400–700 nm) was found to be directly related to FAW damage, increasing with the extent of damage. In contrast, reflectance in the near-infrared zone (700–1050 nm) exhibited an inverse trend due to different gradients of FAW damage in the maize canopy. The present findings revealed that significant wavebands 391–401, 411–440, 524–527, 568–577, 586–587, 597, 602–694, and 703–1050 nm were identified as highly sensitive to FAW damage. Sensitivity analysis via reflectance sensitivity yielded minima at 413 and 689 nm and maxima at 390, 404, and 694 nm, which were attributed as the most responsive wavelengths for FAW damage characterization. These findings concluded that hyperspectral sensing can effectively detect FAW damage in maize fields. Proximal remote sensing offers a promising tool for FAW monitoring, providing a more efficient alternative to manual inspection.