Figure 3.

Frame and leaf image segmentation processing chain. In the first step, the white leaf-holding frames are used to define the regions of interest (ROIs) that contain the leaf fragments. Next, the plate image is split on four ROIs, each one containing a row of leaf samples that typically belong to a single genotype but to a different individual plant (e.g., each leaf fragment in a row is a technical replicate). In parallel, the back-illuminated image of the plate is used to separate the leaves of each other. The final leaf segmentation is done by combining the row and leaf ROIs.