Corn360: a method for quantification of corn kernels

Samantha Gillette, Lu Yin, Penny M.A. Kianian, Wojciech P. Pawlowski, Changbin Chen

Research output: Contribution to journalArticlepeer-review


Background: The rapidly advancing corn breeding field calls for high-throughput methods to phenotype corn kernel traits to estimate yield and to study their genetic inheritance. Most of the existing methods are reliant on sophisticated setup, expertise in statistical models and programming skills for image capturing and analysis. Results: We demonstrated a portable, easily accessible, affordable, panoramic imaging capturing system called Corn360, followed by image analysis using freely available software, to characterize total kernel count and different patterned kernel counts of a corn ear. The software we used did not require programming skills and utilized Artificial Intelligence to train a model and to segment the images of mixed patterned corn ears. For homogeneously patterned corn ears, our results showed accuracies of 93.7% of total kernel count compared to manual counting. Our method allowed to save an average of 3 min 40 s per image. For mixed patterned corn ears, our results showed accuracies of 84.8% or 61.8% of segmented kernel counts. Our method has the potential to greatly decrease counting time per image as the number of images increases. We also demonstrated a case of using Corn360 to count different categories of kernels on a mixed patterned corn ear resulting from a cross of sweet corn and sticky corn and showed that starch:sweet:sticky segregated in a 9:4:3 ratio in its F2 population. Conclusions: The panoramic Corn360 approach enables for a portable low-cost high-throughput kernel quantification. This includes total kernel quantification and quantification of different patterned kernels. This can allow for quick estimate of yield component and for categorization of different patterned kernels to study the inheritance of genes controlling color and texture. We demonstrated that using the samples resulting from a sweet × sticky cross, the starchiness, sweetness and stickiness in this case were controlled by two genes with epistatic effects. Our achieved results indicate Corn360 can be used to effectively quantify corn kernels in a portable and cost-efficient way that is easily accessible with or without programming skills.

Original languageEnglish (US)
Article number23
JournalPlant Methods
Issue number1
StatePublished - Dec 2023


  • Corn
  • High-throughput phenotyping
  • Image analysis
  • Kernel color
  • Kernel texture
  • Low-cost

ASJC Scopus subject areas

  • Biotechnology
  • Genetics
  • Plant Science


Dive into the research topics of 'Corn360: a method for quantification of corn kernels'. Together they form a unique fingerprint.

Cite this