Principle Researcher Korea Institute of Science and Technology
Disclosure: Disclosure information not submitted.
We conducted research to develop image-based measurement approach to effectively obtain growth related phenotype data from greenhouse tomato. Deep learning model and additive algorithm has been prepared to measure following phenotypes, growth length(per week), flower height, stem diameter, the number of flower and fruit per cluster. Each model required over 5,000 image annotation, training and test validation. Measurement accuracy of our approach were compared with tape measure and vernier caliper. Overall variation of measurements were within 10 percent of standard value, but most of variation seems to be caused by differences in camera pose. This work was supported by RDA(Rural Development Administration) project (Project number is PJ013891032019)