*3.4. Optimal Sensor System for Plant Disease Detection*

Two sensor systems were evaluated, both showing strengths and weaknesses. In direct comparison, the flying sensor system had strong advantages in usability, throughput, and commercial viability. The ground sensing system was much more sensitive, as a single symptom with a diameter of a few mm could be recorded. Such a spatial resolution could only be obtained by the flying system at a flight altitude of around 1 m above the canopy. However, with the used system this was not possible, as the downstream from the rotors would have strongly moved the canopy. Alternatively, an optical zoom could be applied, presumably reducing the light flux as well as the throughput of the overall system. This could be compensated for by an increased spatial resolution of the sensing array.

Summarizing, based on the experiments conducted during the presented study, we propose a focus on a UAV flying at low height in combination with a frame-based spectral camera sensing in around 15 equally distributed bands. A tunable band configuration would be an alternative that could use bands optimized for every single disease scenario, e.g., crop species, crop developmental stage, assumed disease setting, assumed symptom maturity. The spatial resolution should be set at around 1 mm GSD, a value that allows the detection small symptoms but neglects the high-frequency noise caused by the complex surface structure of plants [29,51].
