Appendix 7—table 3. Comparison between 2D vs 3D super pixels.
From out experiments, segmentation quality is almost always improved by the usage of 3D super pixels. On the other side, the user should be aware that this improvement comes at the cost of a large slow-down of the pipeline (roughly × 4.5 on our system Intel Xenon E5-2660, RAM 252 Gb).
ds3 confocal + rescaling | ||||||||
---|---|---|---|---|---|---|---|---|
Over/under factor 0.7 | ||||||||
Dataset | Super Pixels 2D (Default) | Super Pixels 3D | ||||||
ARand | VOIsplit | VOImerge | time (s) | ARand | VOIsplit | VOImerge | time (s) | |
Anther | 0.215 | 1.130 | 0.517 | 600 | 0.167 | 0.787 | 0.399 | 2310 |
Filament | 0.159 | 0.899 | 0.350 | 120 | 0.171 | 0.687 | 0.487 | 520 |
Leaf | 0.117 | 0.502 | 0.247 | 800 | 0.080 | 0.308 | 0.220 | 3650 |
Pedicel | 0.339 | 1.148 | 0.894 | 450 | 0.314 | 0.845 | 0.604 | 2120 |
Root | 0.113 | 0.672 | 0.485 | 210 | 0.101 | 0.356 | 0.412 | 920 |
Sepal | 0.346 | 0.926 | 1.211 | 770 | 0.257 | 0.690 | 0.966 | 3420 |
Valve | 0.444 | 0.828 | 1.138 | 530 | 0.300 | 0.494 | 0.875 | 2560 |
Average | 0.248 | 0.872 | 0.691 | 500 | 0.199 | 0.595 | 0.566 | 2210 |