Developed in collaboration with Dr. Brian Russell at Contempo Lab, they’ve successfully detected seedlings, identified parts, and applied pass/fail rules. Challenges like needle visibility and trunk measurements were overcome with a hybrid machine learning AI approach.
Read the technical report here if you want more details.
The next step is implementing the prototype in a working nursery, expand the data set, and refine the model. Designing an in-shed system with cameras, data processing, and a dashboard interface is also on the roadmap.
Source: Forest Growers Research