Field-based high-throughput phenotyping of plant height in sorghum using different sensing technologies

Plant height is an important morphological and developmental phenotype that directly indicates overall plant growth and is widely predictive of final grain yield and biomass. Currently, manually measuring plant height is laborious and has become a bottleneck for genetics and breeding programs. The goal of this research was to evaluate the performance of five different sensing technologies for field-based high throughput plant phenotyping (HTPP) of sorghum [Sorghum bicolor (L.) Moench] height. With this purpose, (1) an ultrasonic sensor, (2) a LIDAR-Lite v2 sensor, (3) a Kinect v2 camera, (4) an imaging array of four high-resolution cameras were evaluated on a ground vehicle platform, and (5) a digital camera was evaluated on an unmanned aerial vehicle platform to obtain the performance baselines to measure the plant height in the field. Plot-level height was extracted by averaging different percentiles of elevation observations within each plot. Measurements were taken on 80 single-row plots of a US × Chinese sorghum recombinant inbred line population. The performance of each sensing technology was also qualitatively evaluated through comparison of device cost, measurement resolution, and ease and efficiency of data analysis.

We found the heights measured by the ultrasonic sensor, the LIDAR-Lite v2 sensor, the Kinect v2 camera, and the imaging array had high correlation with the manual measurements (r ≥ 0.90), while the heights measured by remote imaging had good, but relatively lower correlation to the manual measurements (r = 0.73).

These results confirmed the ability of the proposed methodologies for accurate and efficient HTPP of plant height and can be extended to a range of crops. The evaluation approach discussed here can guide the field-based HTPP research in general.