List your company
ResearchFormerly WizWand

Surface Normal Estimation on NYUD v2

16.25Mean ErrorSAK
15.839618.609821.3824.1502Jun 14, 2026
Updated 4mo ago

Evaluation results

MethodLinks
SAKCategory=Conventional Multi-task Methods2026.0616.25——
TIGER-LBackbone=ViT-L2026.0616.8——
TIGER-BBackbone=ViT-B2026.0616.95——
MLoRECategory=Conventional Multi-task Methods2026.0618.33——
SEMCategory=Conventional Multi-task Methods2026.0618.45——
TaskPrompterCategory=Conventional Multi-task Methods2026.0618.47——
RADIOCategory=Conventional Multi-task Methods2026.0618.49——
TaskExpertCategory=Conventional Multi-task Methods2026.0618.54——
MTMambaCategory=Conventional Multi-task Methods2026.0618.63——
InvPTCategory=Conventional Multi-task Methods2026.0619.04——
DINOv2-ViT-LBackbone=ViT-L2026.0619.23——
MQTransformerCategory=Conventional Multi-task Methods2026.0619.67——
ATRCCategory=Conventional Multi-task Methods2026.0620.18——
MTI-NetCategory=Conventional Multi-task Methods2026.0620.27——
PAD-NetCategory=Conventional Multi-task Methods2026.0620.88——
SAM-ViT-BBackbone=ViT-B2026.0623.59——
DiffusionMTLCategory=Conventional Multi-task Methods2026.0624.75——
OWLv2-ViT-LBackbone=ViT-L2026.0625.2——
CLIP-ViT-LBackbone=ViT-L2026.0626.51——
DiTASKBackbone=Swin-Tiny, Pre-trained=ImageNet-1k, Trainable Parameters (M)=3.062026.03—27.25-10.45
FAARBackbone=Swin-Tiny, Pre-trained=ImageNet-1k, Trainable Parameters (M)=2.85, Rank (rinit)=642026.03—26.35-7.88
MTL - Full Fine TuningBackbone=Swin-Tiny, Pre-trained=ImageNet-1k, Trainable Parameters (M)=28.102026.03—24.33-8.49
MTLoRABackbone=Swin-Tiny, Pre-trained=ImageNet-1k, Trainable Parameters (M)=9.70, Rank (r)=642026.03—29.53-15.3
Single TaskBackbone=Swin-Tiny, Pre-trained=ImageNet-1k, Trainable Parameters (M)=84.002026.03—22.830
TADFormerBackbone=Swin-Tiny, Pre-trained=ImageNet-1k, Trainable Parameters (M)=8.90, Rank (r)=642026.03—27.48-10.42