Fig. 6From: A two-step registration-classification approach to automated segmentation of multimodal images for high-throughput greenhouse plant phenotypingSummary of evaluation of eight different color classification models from Table 2. a Comparison of average model accuracy by reproducing the same training dataset they were trained to crossover different plant species and camera views. b Distribution of accuracy of all 288 case-scenario models considering for different camera views (top/side), plant types (arabidopsis, wheat, maize) and developmental stages by application to test samples. c Cross-evaluation of accuracy of classification models trained with manually annotated images of different plant developmental stages, including juvenile (I), mid-stage (II), adult (III) plants, and their combinations (i.e. IÂ +Â II, IÂ +Â III, IIÂ +Â III, IÂ +Â IIÂ +Â III) by application to the same training set (0), as well as test samples of juvenile (I), mid-stage (II) and adult (III) plantsBack to article page