Full Shape Canonicalization on ShapeNet 13 categories (test)
0.1325CC (Category-Level Consistency)Compass
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CompassAggregation Type=Average across categories, Training Strategy=Per-category Model2022.01 | 0.1325 | 0.0703 | 0.1275 | |
| CompassTraining Strategy=Multi-category Joint Model2022.01 | 0.1283 | 0.0507 | 0.1247 | |
| Canonical Capsules (CaCa)Aggregation Type=Average across categories, Training Strategy=Per-category Model2022.01 | 0.1262 | 0.0698 | 0.087 | |
| Canonical Capsules (CaCa)Training Strategy=Multi-category Joint Model2022.01 | 0.1132 | 0.0395 | 0.106 | |
| Ground TruthAggregation Type=Average across categories2022.01 | 0.1108 | — | — | |
| Ground TruthTraining Strategy=Multi-category Joint Model2022.01 | 0.1108 | — | — | |
| ConDor (F)Method Variant=Full Shape, Training Strategy=Multi-category Joint Model2022.01 | 0.1035 | 0.0394 | 0.0902 | |
| ConDor (F+P)Method Variant=Full+Partial, Aggregation Type=Average across categories, Training Strategy=Per-category Model2022.01 | 0.103 | 0.0314 | 0.0706 | |
| PCAAggregation Type=Average across categories, Training Strategy=Per-category Model2022.01 | 0.1026 | 0.0535 | 0.081 | |
| PCATraining Strategy=Multi-category Joint Model2022.01 | 0.1026 | 0.0535 | 0.081 | |
| ConDor (F)Method Variant=Full Shape, Aggregation Type=Average across categories, Training Strategy=Per-category Model2022.01 | 0.1006 | 0.0308 | 0.0723 | |
| ConDor (F+P)Method Variant=Full+Partial, Training Strategy=Multi-category Joint Model2022.01 | 0.1005 | 0.0329 | 0.0843 |