WISRD Research & Engineering Journal

A splitting function takes each contour cluster (ultimately an individual fragment), assigns it a number, and separates it out into its own png image that is stored in Jupyter. A capture of the individual fragment on the far left is shown in Figure 49.

Figure 49 . Image of a coral fragment that was separated out from the rest of the image. 5. Evaluation of PlantCV Quantifying the rate of coral growth is central to the type of research being conducted by the Mariner Ocean Research Institute and Wildwood Institute for STEM Research and Development. The overarching research question under joint investigation is, “Does the fragment size of corals propagated by microfragmentation influence their growth rate?” No accounts of using PlantCV for measuring the growth of corals have been found. We contacted one of the developers of PlantCV at the Danforth Center about this project, but never received a reply to these questions. PlantCV is a relatively new method of artificial intelligence computer vision for plant phenotyping and while it does have some documentation and a couple of video tutorials, there is no ready-to-go code for our use case. Or direct user interface to operate from. The majority of this internship time was filled with creating the numerous supporting systems necessary to image coral fragments, which are described in Sections 1-3 above. Learning enough python to write the workflow code with the necessary segmentation steps was extremely time consuming. Before the end of the 8 weeks, the primary investigator had hoped to use PlantCV to take area measurements (in pixels) over the workflow, but unfortunately had to stop just short of that. We was able to segment out the fragments, but a little more time would be needed to capture the tips of several fragments. Also, the rulers hindered the segmentation process, but we did not feel it was necessary to fix that at this time since we wanted to also try to separate out the individual fragments within one image. While using PlantCV to separate objects, it became clear that in the next iteration, it is crucial to physically separate the fragments further apart. We found that taking certain specific actions going forward could make using this tool more effective for our coral restoration research:

64