From Seeds to Leaves:

How AI is Revolutionizing Plant Breeding

Nikee Shrestha

2022-2025 FFAR Fellow, University of Nebraska-Lincoln

Lincoln, NE

Plant breeding has come a long way since farmers first began saving seeds from the best plants thousands of years ago. Today, breeders use cutting-edge tools like genomic sequencing to identify genes linked to specific traits of interest. My research focuses on improving how we measure these traits, which is crucial for accurately finding the genes responsible for them.

For example, if we want to find genes associated with plant height, we first collect data from a variety of plants with different heights and genetic profiles. Then, using statistical analysis, we link the genetic diversity to plant height and identify the gene responsible. While this process sounds straightforward for simple and easy-to-measure traits like plant height, it becomes much trickier for more complex traits.

That’s where a part of my research at the University of Nebraska-Lincoln comes in—developing better ways to measure both simple and complex traits. Early efforts focused on matching human measurements using tools like drones to capture traits like plant height. Now, advances in artificial intelligence allow us to measure traits faster and more accurately than humans, even for challenging traits.

What took humans 20–30 minutes per plant, the AI accomplished in minutes, with greater accuracy. This allowed us to identify genes linked to this complex trait. Nikee Shrestha
2022-2025 FFAR Fellow, University of Nebraska-Lincoln

Case Study 1: Unlocking the Secrets of Sorghum Seed Color

One project from my first year as a PhD student highlights this. I used a computer vision model to analyze the color of sorghum seeds from thousands of genotypes. Sorghum seeds vary in color due to bioactive compounds like polyphenols, tannins and carotenoids, which affect human gut health. Traditional methods rely on subjective human assessment for different color shades, which blend continuously. By quantifying seed color based on red, green and blue components, AI models provided consistent, precise measurements that outperformed human evaluations. This high-quality data not only validated known genes linked to seed color but also identified new ones previously missed due to limitations in data resolution.

From (top left) seed scans to (bottom right) seed segment; use of simple flatbed scanner to scan sorghum seeds to get final measurement using computer vision model.
Measuring phyllotaxy by hand (top two images) looks different than measuring phyllotaxy by computer (bottom two images).

Case Study 2: Deciphering the Enigma of Phyllotaxy

AI isn’t just improving existing measurements; it’s enabling us to measure traits humans struggle with. Take phyllotaxy, the arrangement of leaves around a plant’s stem, which influences light and water use. Optimizing phyllotaxy helps plants maximize photosynthesis by ensuring light reaches lower leaves. However, measuring it by hand is time-consuming and inconsistent. In a project led by my lab mate, Jensina Davis, and in collaboration with researchers from Purdue University, we used AI to create 3D plant reconstructions and measure leaf phyllotaxy across hundreds of maize genotypes. What took humans 20–30 minutes per plant, the AI accomplished in minutes, with greater accuracy. This allowed us to identify genes linked to this complex trait.

In the past, researchers aimed to build tools that matched human accuracy. Today, AI surpasses human capabilities, measuring traits faster and with fewer errors. These advancements allow us to focus on connecting traits to genes, paving the way for more efficient plant breeding. From simple traits like seed color to complex ones like phyllotaxy, AI is transforming how we study and improve crops.

The FFAR Fellows: A Transformative Experience

My journey through the FFAR Fellows program these past three years has been so revealing for me and I could not be more grateful for this. Through the program, I’ve had the privilege of being mentored by my amazing mentors which has accelerated my professional growth. I would like to express my deepest gratitude to my advisor Professor James Schnable for

his support throughout this Journey. If I were to choose one word to describe how the FFAR Fellows has impacted my career, it would be “transformative.”  The experiences, skills, and training for soft skills has made me who I am today as a professional individual.

I’m also deeply appreciative of the structure the FFAR Fellows provided, where I connected with my amazing cohort of 2022–2025. This group of talented peers has become more than colleagues—we’ve developed genuine friendships that I will always cherish.

Thank you, FFAR, for empowering me and creating a community that has left an indelible mark on my life and career.