Digital aerial assessment of turfgrass pests for precision management and monitoring epidemics
David McCall: Virginia Tech
<div>Digital image analysis (DIA) drives artificial intelligence that has revolutionized modern society. Applications are increasingly common in the medical, automotive, and agricultural industries. Agricultural remote sensing relies on raw and transformed digital spectral properties. For instance, vegetation indices typically utilize a ratio of visible and near infrared light to discern relative plant phenology. Popularity of aerial imagery collected from drone platforms has increased recently in the turfgrass industry, though practical use of the data is still in its infancy. Large datasets from drone-based DIA comes with unique challenges associated with lighting and spatial alignment that are more controlled with ground measurements. Our lab has developed methods to spatially map pest outbreaks using aerial DIA and computer recognition software. Best data and mosaic alignment is achieved on overcast days that improve contrast and reduce shadows. Glare from sunny days create exposure problems and can be minimized by imaging at solar noon. Mosaicked images should include >50% overlap to provide useful redundancy. Ground reference points with high spatial accuracy will improve georectification of mosaicked images. Digital recognition of ground-validated disease symptoms is possible through computer learning approaches. Geospatial disease maps can reduce pesticide inputs through targeted applications and offer a powerful model for improved epidemiological investigation.</div>
View Presentation |
