quantitative comparison
Representation with Local Geometry Regularization Supplemental Material
We compare our method with four competing methods in Table 1 of the main paper. We also use the score reported by [5]. We found that the corner-based methods, e.g., HEA T and RoomFormer, fail to reconstruct the correct floorplans and are easily affected by the irregular Heat: Holistic edge attention transformer for structured reconstruction. Real-world perception for embodied agents.
- Asia > South Korea > Seoul > Seoul (0.04)
- Europe > Netherlands > North Holland > Amsterdam (0.04)
- Africa > Mali (0.04)
- Research Report > Experimental Study (0.93)
- Research Report > New Finding (0.67)
- North America > United States > Massachusetts > Middlesex County > Cambridge (0.04)
- Asia > China > Guangdong Province > Shenzhen (0.04)
- Research Report > Experimental Study (1.00)
- Research Report > New Finding (0.67)
- Asia > Middle East > Israel (0.05)
- Asia > China > Hong Kong (0.05)
- Asia > China > Guangdong Province > Shenzhen (0.05)
- North America > United States > California > Santa Clara County > Palo Alto (0.05)
- North America > Canada (0.04)
DecomposedKnowledgeDistillationfor Class-IncrementalSemanticSegmentation Supplement
All numbers are obtained by averaging results over five runs with standard deviations. All numbers are obtained by averaging results over five runs with standard deviations. Initialization Ldkd 19-1(2steps) Random Ours Recall Precision 3 3 3 78.88 NFP, andNFN are the number of true positives, false positives, and false negatives, respectively. All numbers are obtained by averaging results over five runs with standard deviations.