Table 6 Parameter studies on varying the label selection thresholds, tested on the Oakland dataset
From: Pseudo-labelling-aided semantic segmentation on sparsely annotated 3D point clouds
tconf | Overall accuracy (%) | Average F-score (%) | Per-class F-scores(%) | ||||
|---|---|---|---|---|---|---|---|
|  |  |  | Foliage | Wire | Pole | Ground | Façade |
0.99 | 92.4 | 58.3 | 82.4 | 19.1 | 8.9 | 98.9 | 82.5 |
0.97 | 94.6 | 73.5 | 87.3 | 43.6 | 45.5 | 97.7 | 93.5 |
0.95 | 96.6 | 74.2 | 92.0 | 40.2 | 46.2 | 99.3 | 93.3 |
0.90 | 94.4 | 73.3 | 87.0 | 40.1 | 48.3 | 97.5 | 93.8 |
0.85 | 94.7 | 74.4 | 87.4 | 41.1 | 52.3 | 97.8 | 93.6 |
0.80 | 96.6 | 75.6 | 91.9 | 41.7 | 48.9 | 99.2 | 93.6 |
0.70 | 94.5 | 74.2 | 86.9 | 39.2 | 53.4 | 97.6 | 93.6 |
0.60 | 94.3 | 76.0 | 86.3 | 48.4 | 54.5 | 97.4 | 93.4 |
Seg-aided [17] | 96.6 | 68.4 | 93.7 | 46.5 | 8.7 | 99.5 | 93.9 |