Surface Normals Estimation on NYU Depth v2

Metric: % < 11.25 (higher is better)

LeaderboardDataset

Loading chart...

Results

Submit a result

Hide extra data

Sort:

#	Model↕	% < 11.25▼	Extra Data	Paper	Date↕	Code
1	Metric3Dv2(L, FT)	68.8	Yes	Metric3Dv2: A Versatile Monocular Geometric Foun...	2024-03-22	Code
2	PolyMaX(ConvNeXt-L)	65.66	No	PolyMaX: General Dense Prediction with Mask Tran...	2023-11-09	Code
3	iDisc	63.8	No	iDisc: Internal Discretization for Monocular Dep...	2023-04-13	Code
4	Bae et al.	62.2	No	Estimating and Exploiting the Aleatoric Uncertai...	2021-09-20	Code
5	Marigold + E2E FT(zero-shot)	61.4	No	Fine-Tuning Image-Conditional Diffusion Models i...	2024-09-17	Code
6	Floors are Flat	59.5	No	Floors are Flat: Leveraging Semantics for Real-T...	2019-06-16	Code

#1Metric3Dv2(L, FT)SOTA
68.8
% < 11.25· Extra Data· 2024-03-22
Metric3Dv2: A Versatile Monocular Geometric Foundation Model for Zero-shot Metric Depth and Surface Normal Estimation Code
#2PolyMaX(ConvNeXt-L)SOTA
65.66
% < 11.25· 2023-11-09
PolyMaX: General Dense Prediction with Mask Transformer Code
#3iDiscSOTA
63.8
% < 11.25· 2023-04-13
iDisc: Internal Discretization for Monocular Depth Estimation Code
#4Bae et al.SOTA
62.2
% < 11.25· 2021-09-20
Estimating and Exploiting the Aleatoric Uncertainty in Surface Normal Estimation Code
#5Marigold + E2E FT(zero-shot)
61.4
% < 11.25· 2024-09-17
Fine-Tuning Image-Conditional Diffusion Models is Easier than You Think Code
#6Floors are FlatSOTA
59.5
% < 11.25· 2019-06-16
Floors are Flat: Leveraging Semantics for Real-Time Surface Normal Prediction Code