Advances, Systems and Applications
Table 1 Comparison of Different Methods on ModelNet40
From: Efficient 3D object recognition in mobile edge environment
Type | Methods | Accuracy |
|---|---|---|
Supervised | LFDÂ [12] | 75.5% |
MVCNNÂ [12] | 89.9% | |
MVCNN (high-quality)Â [59] | 95.5% | |
DGCNNÂ [71] | 92.2% | |
Point-TnTÂ [36] | 92.6% | |
PointStack [37] | 93.3% | |
Unsupervised | Primitive-GANÂ [72] | 86.4% |
FoldingNet(ModelNet40)Â [29] | 86.2% | |
FoldingNet(ShapeNet)Â [29] | 88.4% | |
CrossPoint [32] | 89.1% | |
Semi-supervised | LFDÂ [12] | 60.8% |
FoldingNet [29] | 76.2% | |
OSSSLÂ [40] | 85.5% | |
Co-training [38] | 89.0% | |
Ours | 91.5% |