Semantic Segmentation On Nyu Depth V2

Metriken

Mean IoU

Ergebnisse

Leistungsergebnisse verschiedener Modelle zu diesem Benchmark

Modellname	Mean IoU	Paper Title	Repository
DynMM (ResNet-50)	51.0%	Dynamic Multimodal Fusion
GeminiFusion (Swin-Large)	60.9	GeminiFusion: Efficient Pixel-wise Multimodal Fusion for Vision Transformer
CMX (B2)	54.4%	CMX: Cross-Modal Fusion for RGB-X Semantic Segmentation with Transformers
LS-DeconvNet	45.9%	Locality-Sensitive Deconvolution Networks With Gated Fusion for RGB-D Indoor Semantic Segmentation	-
ESANet (R18-NBt1D )	48.17	Efficient RGB-D Semantic Segmentation for Indoor Scene Analysis	-
3DGNN	43.1%	3D Graph Neural Networks for RGBD Semantic Segmentation
AsymFormer	55.3%	AsymFormer: Asymmetrical Cross-Modal Representation Learning for Mobile Platform Real-Time RGB-D Semantic Segmentation
AsymFusion (ResNet-152)	51.2%	Learning Deep Multimodal Feature Representation with Asymmetric Multi-layer Fusion
SwinMTL	58.14%	SwinMTL: A Shared Architecture for Simultaneous Depth Estimation and Semantic Segmentation from Monocular Camera Images
OmniVec2	63.6	OmniVec2 - A Novel Transformer based Network for Large Scale Multimodal and Multitask Learning	-
MMAF-Net-152	44.8%	Multi-Modal Attention-based Fusion Model for Semantic Segmentation of RGB-Depth Images	-
GeminiFusion (MiT-B3)	56.8	GeminiFusion: Efficient Pixel-wise Multimodal Fusion for Vision Transformer
Cross-stitch	19.3%	Cross-stitch Networks for Multi-task Learning
HN-network	33.49%	RGB-based Semantic Segmentation Using Self-Supervised Depth Pre-Training	-
CMNeXt (B4)	56.9%	Delivering Arbitrary-Modal Semantic Segmentation
CFN	47.7%	Cascaded Feature Network for Semantic Segmentation of RGB-D Images	-
NDDR-CNN	43.3%	NDDR-CNN: Layerwise Feature Fusing in Multi-Task CNNs by Neural Discriminative Dimensionality Reduction
ACNet	48.3%	ACNet: Attention Based Network to Exploit Complementary Features for RGBD Semantic Segmentation
Malleable 2.5D (ResNet-101)	50.9%	Malleable 2.5D Convolution: Learning Receptive Fields along the Depth-axis for RGB-D Scene Parsing
SGACNet (R34-NBt1D)	49.4%	Spatial-information Guided Adaptive Context-aware Network for Efficient RGB-D Semantic Segmentation

0 of 116 row(s) selected.