Monocular Depth Estimation On Nyu Depth V2

Metriken

Delta u003c 1.25

Delta u003c 1.25^2

Delta u003c 1.25^3

RMSE

absolute relative error

log 10

Ergebnisse

Leistungsergebnisse verschiedener Modelle zu diesem Benchmark

Modellname	Delta u003c 1.25	Delta u003c 1.25^2	Delta u003c 1.25^3	RMSE	absolute relative error	log 10	Paper Title	Repository
UniK3D (FT, metric)	0.989	0.998	1.000	0.173	0.044	0.019	-	-
GLPDepth	0.915	0.988	0.997	0.344	0.098	0.042	Global-Local Path Networks for Monocular Depth Estimation with Vertical CutDepth
Depth-Map-Decomposition	0.907	0.986	0.997	0.362	0.100	0.043	Depth Map Decomposition for Monocular Depth Estimation
SOM	-	-	-	0.604	-	-	Structure-Attentioned Memory Network for Monocular Depth Estimation	-
DMD	0.953	0.989	0.996	0.296	0.072	0.031	Zero-Shot Metric Depth with a Field-of-View Conditioned Diffusion Model	-
PrimeDepth + Depth Anything	0.977	-	-	-	0.046	-	PrimeDepth: Efficient Monocular Depth Estimation with a Stable Diffusion Preimage
HybridDepth	0.988	1.000	1.000	0.128	0.026	-	HybridDepth: Robust Metric Depth Fusion by Leveraging Depth from Focus and Single-Image Priors
SC-DepthV2	0.820	0.956	-	0.532	0.138	0.059	Auto-Rectify Network for Unsupervised Indoor Depth Estimation
P3Depth	0.898	0.981	0.996	0.356	0.104	0.043	P3Depth: Monocular Depth Estimation with a Piecewise Planarity Prior
DPT-Hybrid	0.904	0.988	0.994	0.357	0.110	0.045	Vision Transformers for Dense Prediction
Metric3Dv2(L, FT)	0.989	0.998	1.000	0.183	0.047	0.020	Metric3Dv2: A Versatile Monocular Geometric Foundation Model for Zero-shot Metric Depth and Surface Normal Estimation
DINOv2 (ViT-g/14 frozen, w/ DPT decoder)	0.9497	0.996	0.9994	0.279	0.0907	0.0371	DINOv2: Learning Robust Visual Features without Supervision
LocalBins	0.91	0.986	0.997	0.351	0.098	0.042	LocalBins: Improving Depth Estimation by Learning Local Distributions	-
FutureDepth	0.981	0.996	0.999	0.233	0.063	0.027	FutureDepth: Learning to Predict the Future Improves Video Depth Estimation	-
DSN	-	-	-	0.429	-	-	On Deep Learning Techniques to Boost Monocular Depth Estimation for Autonomous Navigation	-
Li et al.	-	-	-	0.635	-	-	A Two-Streamed Network for Estimating Fine-Scaled Depth Maps from Single RGB Images	-
Metric3D (ConvNeXt-Large, Zero-shot testing)	0.944	0.986	0.995	0.310	0.083	0.035	Metric3D: Towards Zero-shot Metric 3D Prediction from A Single Image
Depth-Map-Decomposition-HRWSI	0.913	0.987	0.998	0.355	0.098	0.042	Depth Map Decomposition for Monocular Depth Estimation
NDDepth	0.936	0.991	0.998	0.311	0.087	0.038	NDDepth: Normal-Distance Assisted Monocular Depth Estimation	-
UniDepth (Zero-shot)	0.984	0.997	0.999	0.201	0.058	0.024	UniDepth: Universal Monocular Metric Depth Estimation

0 of 85 row(s) selected.