Table 4 Error rate (%) of the proposed and other stereo matching algorithms
From: Shape-reserved stereo matching with segment-based cost aggregation and dual-path refinement
Methods | AdaStereo [38] | tMGM [39] | iResNet [40] | DeepPruner [41] | Proposed | |
|---|---|---|---|---|---|---|
Adirondack | Non-occ | 28.1 | 29.9 | 25.1 | 36.9 | 16.6 |
ArtL | Non-occ | 11.0 | 11.9 | 20.2 | 37.4 | 30.9 |
Jadeplant | Non-occ | 44.4 | 28.4 | 47.9 | 56.5 | 21.4 |
Motorcycle | Non-occ | 37.0 | 16.1 | 35.6 | 41.2 | 11.5 |
MotorcycleE | Non-occ | 37.9 | 16.1 | 36.2 | 41.7 | 35.5 |
Piano | Non-occ | 39.5 | 25.8 | 37.4 | 37.9 | 14.2 |
PianoL | Non-occ | 62.2 | 36.3 | 59.8 | 47.4 | 43.2 |
Pipes | Non-occ | 36.7 | 14.2 | 34.5 | 47.5 | 16.3 |
Playroom | Non-occ | 49.8 | 38.5 | 46.1 | 52.9 | 34.4 |
Playtable | Non-occ | 51.4 | 30.4 | 40.2 | 44.9 | 21.7 |
PlaytableP | Non-occ | 37.6 | 25.4 | 25.3 | 38.6 | 15.3 |
Recycle | Non-occ | 31.0 | 31.0 | 35.0 | 34.9 | 10.0 |
Shelves | Non-occ | 63.8 | 52.7 | 59.6 | 52.7 | 15.5 |
Teddy | Non-occ | 14.5 | 11.7 | 15.0 | 22.9 | 11.9 |
Vintage | Non-occ | 51.8 | 52.8 | 47.7 | 41.7 | 30.1 |
Average | 36.6 | 25.3 | 35.1 | 41.2 | 21.9 | |