Advanced data technologies are poised to revolutionize farming in a way that enriches the world with greater food security and sustainability.

With collaboration between Texas A&M AgriLife Research and Extension and Lyles School of Civil Engineering at Purdue University, Drs. Juan Landivar and Jinha Jung’s Digital Agriculture team is at the forefront of combining age-old agricultural wisdom with state-of-the-art methods to help farmers grow more food and minimize damage from storms like the one that devastated Texas in February.

The multidisciplinary team of crop experts, computer scientists, and geomatics engineers launched their work to build UASHub, a web-based system that compiles and evaluates data from Unmanned Aircraft Systems (UAS), better known as drones, in 2015.

Recognizing the enormous computing resources and flexibility needed to make the system—one that manages 120 TB of data accumulated between 2015 and 2020—available to research scientists within the Texas A&M System, the team migrated to Oracle Cloud Infrastructure in August 2020 through the Oracle for Research program.

Now, “Oracle Cloud is right at the center of our digital agriculture program,” Landivar said.

Working in collaboration with Purdue University, the Texas team is ready to expand access to its groundbreaking project to other agriculture researchers across the U.S.—and eventually commercial partners.

UASHub has already been rolled out across Texas A&M AgriLife’s 11 research and extension centers, each one addressing the specific agricultural needs of their region of Texas.

The system stores, manages, processes, and visualizes massive data sets, applying advanced machine learning algorithms to gain insights from the arrays of photos drones capture while flying over acres of fields. Artificial intelligence assesses every square meter to let farmers know when to water or fertilize their crops or the best ways to control disease.

“It’s opening the door to opportunities we didn’t have three or five years ago,” Landivar said.

Take the February storms that caused some $600 million in agricultural losses across Texas.

“This type of research will provide quick and easy assessment of the damage by the recent storm and freeze,” Landivar said. “The system will speed up how we visualize UAS images, assess storm damage, and plan the next step for agriculture fields.”

The UAS data processing employs Structure from Motion (SfM) analysis, a cutting-edge technique where 3-dimensional models are rendered from a series of overlapping 2-dimensional images.

By adding depth, the 3D models allow several crop metrics to be calculated in a way that wouldn’t be possible from traditional remote sensing platforms.

But SfM isn’t easy—rendering in 3D is extremely computationally intensive, requiring clusters of GPUs running in parallel to deliver timely results. Analyzing the terabytes of data involved in SfM on-premises requires a capital investment prohibitive to small academic institutions.

While the primary goal of the Texas A&M and Purdue researchers in adopting cloud was realizing cost and construction flexibility, they also got the benefit of superior performance and efficiency.

The team benchmarked and compared SfM analysis executed on-premises and on OCI. With the right virtual machine shape and storage architecture, they saw as much as a 35 percent performance gain with OCI.

Oracle Global Senior Solutions Architect Rajib Ghosh worked with the group to optimize that storage architecture and select the best bare-metal shapes for the compute- and I/O-intensive workloads.

“It was a pleasure to help these groundbreaking scientists get game-changing benefits from Oracle Cloud that will enable them to expand the scope of the important work they are doing,” Ghosh said. “With Oracle Cloud Infrastructure to power their uniquely demanding workloads, the team is positioned to scale their work and make it accessible to more academic and industry partners.”

Alison Derbenwick Miller, vice president for Oracle for Research, said the project epitomizes Oracle’s mission to help the world’s leading researchers access flexible and powerful computing resources to propel breakthroughs that benefit all.

“The Texas A&M and Purdue scientists are doing work that will help solve one of the world’s most-pressing and intractable problems. We’re proud to make it easier for them to not only help make farmers more financially secure, but also deliver a more stable and reliable food supply to a world dealing with increasingly complex environmental challenges,” Derbenwick Miller said.

Dr. Jinha Jung, an assistant professor in the Lyles School of Civil Engineering at Purdue who specializes in Geomatics Engineering, said the system will go far in helping meet the needs of the world in 10 to 20 years, when resources will likely be more constrained.

When Jung and Landivar initially envisioned the system, it didn’t matter to their teams whether the prototype was built on-premises or in the cloud, Jung said.

But expanding its reach and scale was another matter.

“Cloud gives us the flexibility to increase instances more efficiently without worrying about how to maintain security and scale-up computational resources if needed,” Jung said.

Those same AI and big data techniques might ultimately prove even more revolutionary in enhancing farm productivity once they are merged with advanced phenotyping tools that employ cutting-edge genetics, Jung added.

There are plenty of breakthroughs to come now that the UASHub is on the cusp of being turned into a business that will help food producers across the U.S. and the world.

“We’re working on how to duplicate this platform for other companies,” Jung said. “More and more will take advantage of this kind of technology.”

 

By Joe Tsidulk
Source Oracle Cloud

Previous Simplifying SaaS, PaaS, And IaaS
Next Improve Your App Mileage With Android For Cars App Library