Skip to main content
. 2020 Jul 2;20(13):3710. doi: 10.3390/s20133710

Figure 6.

Figure 6

Results of the combined segmentation. On the left the original input image and below the ground truth label are shown. The top row depicts a two-dimensional embedding space. The bottom row depicts the corresponding binary segmentation and the assignment of the clusters. The snapshots are created after 1, 2, 3, 10 and 80 gradient updates. The network was trained solely on the shown input image to generate the results shown here for illustration purposes.