b'PARTNER CONTENTTransforming Crop Breeding with Trait Extraction from Drone ImageryBy: Lee WestT rait extraction from drone imagery is a transformative tool for crop breed-ers, offering unprecedented insights into crop performance and health. This technology not only expedites the breed-ing process but also equips breeders with the knowledge needed to develop crops that can withstand the challenges of 21st century agriculture. Until recently, assessing plant traits in-volved time-consuming and labor-intensive fieldwork. Drone imagery, however, acceler-ates this process by providing high-resolu-tion data on a large scale. From plant height and canopy cover to chlorophyll content and stress indicators, drones can capture a wealth of information that is instrumental to understanding a crops performance.Plant Trait Extraction: A Game-Changer for Crop BreedersIn 1973, John Rouse first applied a concept of using a reflectance/absorption ratio ofRecent development in machine learn- protocols, addressing privacy concerns, two spectral bands (red and near-infrared)ing allows for classification and quan- and refining data analysis algorithms are to differentiate between different ecosys- tification of features in drone images.ongoing endeavors. Additionally, integrat-tems from satellite imagery. That was theThis development led to traits like standing this technology into the broader agri-birth of Normalized Difference Vegetationcounts, ground coverage, and the countingcultural landscape requires collaboration Index (NDVI).Over the years, plant scien- of organs like wheat heads and sorghumamong researchers, breeders, and technol-tists have devised many different ratios ofpanicles. ogy developers.absorbed and reflected light and combinedDrones can now be equipped with them into an alphabet soup (MCARI,thermal sensors which can detect earlyData Driven Decision MakingNDRE, NRI, LAI, MSAVI etc.) of vegetativesignals of plant disease. LiDAR can also beThe sheer volume of data generated by indices to measure phenotypic and pheno- attached as payloads on drones, increasingdrone imagery necessitates sophisticated logical characteristics of plants.the accuracy and resolution of 3-D traits.data analysis techniques. Artificial intel-In the early aughts, image stitchingFor many of the key crops, these traitsligence and machine learning algorithms process began to evolve and when appliedare in production mode via leading imag- come into play, processing large datasets to drone image acquisition, allowed for theing solutions organizations like Hiphen.Asto extract meaningful insights. This data-building of 3D recreations of fields. Thisthese traits become more routine, the datadriven approach empowers crop breed-enables the extraction of architecturalgenerated is being analyzed through timeers to make informed decisions, identify traits like plant height and biovolume fromseries or algorithms that combine differ- genetic markers associated with desir-drone image sets. The orthographic projec- ent traits to derive key performance andable traits, and streamline the selection tion from this process also facilitated themanagement markers such as maturity,of promising plant varieties. In the next georeferencing of plot maps, thus open- flowering date and even yield predictions. installment of this column, Hiphen CEO ing the door for extraction of traits fromWhile plant trait extraction from droneAlexis Comar will talk about making sense the small plot work ubiquitous with cropimagery holds immense promise, challeng- of all of this new data that has become researchers and breeders. es remain. Standardizing data collectionavailable. FEBRUARY 2024SEEDWORLD.COM /47'