Skip to main content
Fig. 2 | Plant Methods

Fig. 2

From: Semantic segmentation of plant roots from RGB (mini-) rhizotron images—generalisation potential and false positives of established methods and advanced deep-learning models

Fig. 2

Masks of prediction examples on the ‘unseen’ test data of the Cichorium intybus (RootPainter) dataset; models e-i were trained on the Mixed dataset. Original image from a rhizotron a, manually labelled mask b, and masks derived using the techniques/models: Frangi Vesselness c, Adaptive thresholding d, Support Vector Machine (SVM) e, SegRoot f, UNetGNRes g, U-Net SE-ResNeXt-101 (32 × 4d) h, and U-Net EfficientNet-b6 i. Only the best DL models (f–i; Table 3), i.e., trained with augmented data (+ aug), are displayed; see Table 4, Additional file 2 and text for details

Back to article page

Plant Methods

ISSN: 1746-4811