Skip to main content
. 2015 May 6;16:143. doi: 10.1186/s12859-015-0560-x

Figure 4.

Figure 4

Evaluation of the segmentation results by comparison to ground truth data. A set of 22 images (A) was used to compare manually labeled ground truth images (B) to the computed segmentation results (C). The confusion matrix (D) shows that the proposed method correctly classifies the majority of pixels as ‘leaf’ (Lv), ‘stem’ (St) or ‘background’ (Bg).