HyperAI超神経

プラットフォーム

Point Cloud Registration On 3Dmatch At Least 1

評価指標

RE (all)

Recall (0.3m, 15 degrees)

TE (all)

評価結果

このベンチマークにおける各モデルのパフォーマンス結果

				Paper Title
GeoTransformer	-	95	-	Geometric Transformer for Fast and Robust Point Cloud Registration
NgeNet	4.932	95.0	0.155	Leveraging Inlier Correspondences Proportion for Point Cloud Registration
PCAM-Sparse (All post-processing)	8.9	92.4	0.23	PCAM: Product of Cross-Attention Matrices for Rigid Registration of Point Clouds
PCAM-Soft (All post-processing)	9.8	91.3	0.24	PCAM: Product of Cross-Attention Matrices for Rigid Registration of Point Clouds
DGR (RE (all), TE(all) are reported in PCAM)	9.5	91.3	0.25	Deep Global Registration
Exhaustive Grid Search	-	84.11	-	Addressing the generalization of 3D registration methods with a featureless baseline and an unbiased benchmark
RANSAC-2M	-	66.1	-	Fast Point Feature Histograms (FPFH) for 3D Registration
FGR	-	42.7	-	Fast Global Registration
Go-ICP	-	22.9	-	Go-ICP: A Globally Optimal Solution to 3D ICP Point-Set Registration
Super4PCS	-	21.6	-	Super 4PCS Fast Global Pointcloud Registration via Smart Indexing
ICP (P2Plane)	-	6.59	-	Open3D: A Modern Library for 3D Data Processing
ICP (P2Point)	-	6.04	-	Open3D: A Modern Library for 3D Data Processing
DCP	-	3.22	-	Deep Closest Point: Learning Representations for Point Cloud Registration
PointNetLK	-	1.61	-	PointNetLK: Robust & Efficient Point Cloud Registration using PointNet

0 of 14 row(s) selected.

Point Cloud Registration On 3Dmatch At Least 1 | SOTA | HyperAI超神経