Table 1 Comparison of performances on the NYU dataset.

Method	Venue	Backbone	\(\delta _1\uparrow\)	\(\delta _2\uparrow\)	\(\delta _3\uparrow\)	AbsRel\(\downarrow\)	RMSE\(\downarrow\)	log10\(\downarrow\)	Params\(\downarrow\)
DORN¹⁷	CVPR 2018	ResNet-101	0.828	0.965	0.992	0.115	0.509	0.051	–
BTS⁵⁸	Arxiv 2019	ResNext-101	0.885	0.978	0.994	0.110	0.392	0.047	47.0M
PWA⁵⁹	AAAI 2021	DenseNet161	0.892	0.985	0.997	0.105	0.374	0.045	–
AdaBins¹²	CVPR 2021	EfficientNet-B5	0.903	0.984	0.997	0.103	0.364	0.044	78.0 M
P3Depth⁶⁰	CVPR 2022	ResNet101	0.898	0.981	0.996	0.104	0.364	0.043	94.2M
NeWCRFs¹⁹	CVPR 2022	Swin-L	0.922	0.992	0.998	0.095	0.334	0.041	270.5M
LifelongDepth²⁰	TNNLS 2023	ResNet-34	0.857	0.972	0.993	0.121	0.429	0.052	22.23M
DepthFormer⁶¹	MIR 2023	Swin-L+R-50-C1	0.923	0.989	0.997	0.094	0.329	0.040	273.0M
IEBins¹⁸	NeurIPS 2023	Swin-T	0.893	0.984	0.996	0.108	0.375	0.046	90.7M
TrapAttention⁶²	CVPR 2023	XCiT-M24	0.925	0.988	0.997	0.092	0.332	0.040	94.2M
MDEUncertainty⁶³	TCSVT 2024	Swin-L	0.879	0.977	0.994	0.112	0.420	0.048	–
ASNDepth²²	TPAMI 2024	HRNet-18	0.906	0.985	0.997	0.101	0.377	0.044	–
Metric3Dv2²⁹	ECCV 2024	ConvNeXt-L	0.925	0.983	0.994	0.092	0.341	0.040	203.24M
SimMDE(Ours)	Ours	MSCAN-B	0.925	0.990	0.997	0.091	0.331	0.039	30.9M

The reported numbers are from the corresponding original papers. Best results are marked bold. The second results are marked underlined.

Quick links

Search