Robust 3D Semantic Segmentation On

Métriques

mean Corruption Error (mCE)

Résultats

Résultats de performance de divers modèles sur ce benchmark

Nom du modèle	mean Corruption Error (mCE)	Paper Title	Repository
PolarNet	118.56%	PolarNet: An Improved Grid Representation for Online LiDAR Point Clouds Semantic Segmentation
SqueezeSegV2 (64x2048)	152.45%	SqueezeSegV2: Improved Model Structure and Unsupervised Domain Adaptation for Road-Object Segmentation from a LiDAR Point Cloud
KPConv	99.54%	KPConv: Flexible and Deformable Convolution for Point Clouds
2DPASS	106.14%	2DPASS: 2D Priors Assisted Semantic Segmentation on LiDAR Point Clouds
MinkUNet-18	100.00%	4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks
PIDS-1.2x	104.13%	PIDS: Joint Point Interaction-Dimension Search for 3D Point Cloud	-
Cylinder3D (torchsparse)	103.13%	Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentation
CPGNet	107.34%	CPGNet: Cascade Point-Grid Fusion Network for Real-Time LiDAR Semantic Segmentation
PIDS-2.0x	101.20%	PIDS: Joint Point Interaction-Dimension Search for 3D Point Cloud	-
GFNet	108.68%	GFNet: Geometric Flow Network for 3D Point Cloud Semantic Segmentation
SPVCNN-34	99.16%	Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution
CENet (64x2048)	103.41%	CENet: Toward Concise and Efficient LiDAR Semantic Segmentation for Autonomous Driving
RPVNet	111.74%	RPVNet: A Deep and Efficient Range-Point-Voxel Fusion Network for LiDAR Point Cloud Segmentation	-
WaffleIron	109.54%	Using a Waffle Iron for Automotive Point Cloud Semantic Segmentation
FIDNet (64x2048)	113.81%	FIDNet: LiDAR Point Cloud Semantic Segmentation with Fully Interpolation Decoding
Cylinder3D (spconv)	103.25%	Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentation
MinkUNet-34	100.61%	4D Spatio-Temporal ConvNets: Minkowski Convolutional Neural Networks
SalsaNext (64x2048)	116.14%	SalsaNext: Fast, Uncertainty-aware Semantic Segmentation of LiDAR Point Clouds for Autonomous Driving
SqueezeSeg (64x2048)	164.87%	SqueezeSeg: Convolutional Neural Nets with Recurrent CRF for Real-Time Road-Object Segmentation from 3D LiDAR Point Cloud
RangeNet-21 (64x2048)	136.33%	RangeNet++: Fast and Accurate LiDAR Semantic Segmentation

0 of 22 row(s) selected.