Fig. 3
From: InSegtCone: interactive segmentation of crystalline cones in compound eyes
Performance of the auto-segmentation method in Apis mellifera (green), Bombus terrestris (orange) and Pieris napi (purple). The surface modelling error (a), segmentation time (b), percentage of auto-segmented cones (c) and angular discrepancy between manual and auto-segmentation (d), are represented as a function of the number of eye subregions used in the segmentation algorithm. When drawn, error bars represent the standard error around the mean values