Depth Anything in $360^\circ$: Towards Scale Invariance in the Wild

Hualie Jiang; Ziyang Song; Zhiqiang Lou; Rui Xu; Minglang Tan

cs.CVDec 28, 2025

Depth Anything in $360^\circ$ : Towards Scale Invariance in the Wild

Authors: Hualie Jiang, Ziyang Song, Zhiqiang Lou, Rui Xu, Minglang Tan

Organizations: Insta360 Research

Abstract

Panoramic depth estimation captures the complete 360 $^\circ$ scene geometry, being essential for robotics and AR/VR applications. While perspective depth models have achieved remarkable zero-shot generalization via large-scale training, panoramic methods lag behind, especially for open-world scenes, due to data scarcity. To bridge this gap, we introduce DA360, a panoramic-adapted version of Depth Anything V2. Our key insight is that the base DAV2 model, trained on perspective images to predict affine-invariant disparity, already exhibits good zero-shot performance on panoramas. Building on this, we design a lightweight adaptation framework that (i) learns a per-image shift from the ViT class token with scale-invariant supervision, transforming affine-invariant disparity into scale-invariant disparity that directly yields well-formed 3D point clouds, and (ii) integrates circular padding into the DPT decoder to eliminate seam artifacts, ensuring spatial coherence. Fine-tuned on a combination of synthetic indoor and outdoor panoramic data, DA360 is evaluated on standard real-world indoor benchmarks and our newly curated outdoor dataset, Metropolis. Results show that DA360 not only outperforms the original DAV2 by over 50% and 12% relative error reduction indoors and outdoors, but also surpasses prior specialized methods like PanDA by about 25--35% across all tests, establishing state-of-the-art zero-shot panoramic depth estimation.

Figures & tables

Panel of Figure 3 — Figure 3: Comparisons with SOTA zero-shot monocular depth models on outdoor panoramic images of SUN360.

Figure 4: Our framework is a generalizable panoramic depth estimation model, which fine-tunes a zero-shot pinhole depth model with synthetic panorama depth datasets to produce scale-invariant and boundary-consistent panoramic depth maps.

Models	Backbone	Matterport3D		Metropolis
Models	Backbone	AbsRel $\downarrow$	$\delta_{1}$ $\uparrow$	AbsRel $\downarrow$	$\delta_{1}$ $\uparrow$
Marigold-v1.1 ( Ke et al. 2025 )	SD 2.0	0.2097	65.11	0.7357	26.87
DAV2 ( Yang et al. 2024b )	ViT-S	0.2032	64.56	0.2169	66.34
DAV2-ind ( Yang et al. 2024b )		0.2170	63.98	0.5667	24.89
DAV2-out ( Yang et al. 2024b )		0.2695	54.18	0.4274	34.71
DAV2 ( Yang et al. 2024b )	ViT-B	0.1966	66.03	0.2061	67.64

Table 1: Evaluating pinhole models on panorama datasets.

Figure 5: Circular padding for equirectangular representation, divided into two phases of vertical and horizontal directions.

Models	Matterport3D					Stanford2D3D					Metropolis
Models	AbsRel $\downarrow$	MAE $\downarrow$	RMSE $\downarrow$	RMSE log $\downarrow$	$\delta_{1}\uparrow$	AbsRel $\downarrow$	MAE $\downarrow$	RMSE $\downarrow$	RMSE log $\downarrow$	$\delta_{1}\uparrow$	AbsRel $\downarrow$	MAE $\downarrow$	RMSE $\downarrow$	RMSE log $\downarrow$	$\delta_{1}\uparrow$
DreamCube ( Huang et al. 2025 )	0.2736	0.5201	0.7258	0.1364	56.60	0.2592	0.4130	0.5773	0.1282	61.24	0.4928	15.872	20.682	0.3036	36.16
DAC (L) ( Guo et al. 2025 )	0.1442	0.3185	0.5013	0.0806	82.52	0.1261	0.2166	0.3495	0.0718	86.23	0.5524	17.068	21.748	0.2337	34.09
UniK3D (L) ( Piccinelli et al. 2025 )	0.1046	0.2133	0.3906	0.0645	92.17	0.1011	0.1584	0.2706	0.0607	91.88	0.4842	15.687	20.847	0.2170	33.82
DA 2 (L) ( Li et al. 2025 )	0.1032	0.2256	0.4207	0.0658	89.49	0.0688	0.1165	0.2508	0.0485	95.55	0.3895	11.658	16.131	0.1833	45.82
DAV2 (S) ( Yang et al. 2024b )	0.2032	0.4553	0.7566	0.1128	64.56	0.2561	0.4907	0.7906	0.1416	55.74	0.2169	9.4807	16.057	0.1426	66.34

Table 2: Quantitative comparison. DA360 and DA 2 are evaluated with scale alignment, while all other methods use affine alignment. The postfixes (S), (B), and (L) denote the backbone scales, i.e., ViT-Small, ViT-Base, and ViT-Large, respectively.

Figure 6: Analysis of the curated Metropolis test set. (a) Camera positions colored by elevation, showing 2.2 km $\times$ 1.9 km spatial coverage in Detroit. (b) Histogram of semantic category counts per image; the average is 24 and the range is $[14,36]$ , indicating varying scene complexity.

Panel of Figure 8 — Figure 8: Zero-shot qualitative comparison on Matterport3D ( Chang et al. 2017 ) , Stanford2D3D ( Armeni et al. 2017 ) , Metropolis ( Mapillary 2024 ) , and SUN360 ( Xiao et al. 2012 ) .

Model	Learnable	Learnable	Circular	Training	Loss	Evaluation	Structured3D		Deep360		Matterport3D		Metropolis
Names	Shift	Scale	Padding	Datasets	Loss	Alignment	AbsRel $\downarrow$	$\delta_{1}\uparrow$	AbsRel $\downarrow$	$\delta_{1}\uparrow$	AbsRel $\downarrow$	$\delta_{1}\uparrow$	AbsRel $\downarrow$	$\delta_{1}\uparrow$
SI	✓			S+T	SI	scale	0.0263	99.07	0.0879	89.34	0.1004	91.02	0.1966	72.78
NoShift				S+T	SI	scale	0.0258	99.10	0.1975	84.90	0.1050	90.89	0.4443	60.67
Affine				S+T	SSI	scale+shift	0.0234	99.16	0.1849	82.71	0.0983	90.30	0.2959	55.59
Absol	✓	✓		S+T	Abs	-	0.0817	95.21	0.3319	60.50	0.2880	60.82	0.3409	37.32
CirPad	✓		✓	S+T	SI	scale	0.0258	99.10	0.0838	89.89	0.1007	90.94	0.1934	73.35

Table 3: Ablation study. Loss: SI—scale-invariant (Eq. 3 ), SSI—scale-and-shift-invariant (MiDaS affine-invariant loss ( Ranftl et al. 2020 ) ), Abs—L1. Datasets: S/D/T—Structured3D/Deep360/TartanAir v2.

Panel of Figure 10 — Figure 10: Effectiveness of shift learning and loss configuration.

Panel of Figure 11 — Figure 11: Comparison of w/o and w/ circular padding.

Models	Structured3D	Deep360	Matterport3D	Metropolis
w/o circular padding	0.0226	0.0958	0.1153	0.2661
w/ circular padding	0.0210	0.0871	0.1132	0.2273

Table 4: AbsRel on a 5-pixel ERP boundary margin.

Panel of Figure 12 — Figure S3: Typical problematic samples in Metropolis: a) depth value overflow due to encoding limitations; b) overly sparse valid depth; c) apparent depth errors. Val: ratio of valid depth points; Min: minimum depth value; Max: maximum depth value.

Panel of Figure 13 — Figure S5: Six examples of the curated Metropolis test set.

Appendix figures & tables17 assets

Supplementary material from the paper’s appendix.

Appendix

Figure S6: Comparison of disparity distributions between TartanAir V2 ( Patel et al. 2025 ) and Structured3D ( Zheng et al. 2020 ) : (a) histogram of disparity map median values; (b) histogram of disparity map median absolute deviation (MAD) values; (c) and (d) distributions of median and MAD computed from 5,000 randomly selected disparity maps from TartanAir V2 and Structured3D, respectively.

Panel of Figure 15 — Figure S8: Visual comparisons of different model parameter initialization strategies: ( Random ), DINOv2 , and DAV2 (corresponding to the model SI in the main paper).

Model	Pretrained Models	Structured3D			Deep360			Matterport3D			Metropolis
Names	Pretrained Models	AbsRel $\downarrow$	RMSE $\downarrow$	$\delta_{1}$ $\uparrow$	AbsRel $\downarrow$	RMSE $\downarrow$	$\delta_{1}$ $\uparrow$	AbsRel $\downarrow$	RMSE $\downarrow$	$\delta_{1}$ $\uparrow$	AbsRel $\downarrow$	RMSE $\downarrow$	$\delta_{1}$ $\uparrow$
DAV2	Depth Anything V2 ( Yang et al. 2024b )	0.0263	0.0988	99.07	0.0879	6.2732	89.34	0.1004	0.5047	91.02	0.1966	14.587	72.78
DINOv2	DINOv2 ( Oquab et al. 2023 )	0.0348	0.1218	98.36	0.0967	6.9283	87.60	0.1158	0.5611	88.09	0.2421	17.059	62.97
Random	None/Random	0.1444	0.3278	82.01	0.5039	29.6632	37.06	0.3000	1.0870	50.83	0.5266	17.675	32.57

Appendix

Table S1: Ablation study on different model parameter initialization strategies.

Models	Pre-trained	Finetuning	Evaluation	Matterport3D			Stanford2D3D			Metropolis
Models	Models	Methods	Alignment	AbsRel $\downarrow$	RMSE $\downarrow$	$\delta_{1}\uparrow$	AbsRel $\downarrow$	RMSE $\downarrow$	$\delta_{1}\uparrow$	AbsRel $\downarrow$	RMSE $\downarrow$	$\delta_{1}\uparrow$
PanDA(S)*	DAV2-ind	PanDA	scale+shift	0.1346	0.5090	83.59	0.0957	0.3261	90.80	0.2898	15.037	57.52
PanDA(S) †	DAV2-out	PanDA	scale+shift	0.1414	0.5328	83.26	0.0926	0.3400	92.87	0.2994	15.362	56.90
PanDA(S) ∘	DAV2	PanDA	scale+shift	0.1319	0.5028	85.20	0.0930	0.3293	91.75	0.2517	14.197	64.15
DA360(S)*	DAV2-ind	DA360	scale	0.1076	0.5780	89.48	0.0754	0.3569	93.95	0.1990	14.673	72.77
DA360(S) †	DAV2-out	DA360	scale	0.1059	0.5358	90.33	0.0739	0.3210	94.73	0.2465	15.065	67.02

Appendix

Table S2: Different pre-trained models and finetuning methods.

Models	Matterport3D			Metropolis
Models	1 $\times$	2 $\times$	1 $\times\rightarrow$ 2 $\times$	1 $\times$	2 $\times$	1 $\times\rightarrow$ 2 $\times$
DA360 (S)	0.0963	0.1259	0.0948	0.1922	0.2332	0.1919
DA360 (B)	0.0877	0.1209	0.0861	0.1798	0.2269	0.1783
DA360 (L)	0.0793	0.1075	0.0783	0.1630	0.2054	0.1622

Appendix

Table S3:

AbsRel

at different resolution settings.

Models	Inference	Run	Evaluation	Matterport3D			Stanford2D3D			Metropolis
Models	Style	Time	Alignment	AbsRel $\downarrow$	RMSE $\downarrow$	$\delta_{1}\uparrow$	AbsRel $\downarrow$	RMSE $\downarrow$	$\delta_{1}\uparrow$	AbsRel $\downarrow$	RMSE $\downarrow$	$\delta_{1}\uparrow$
MoGe-2(L)	end2end	0.101s	scale+shift	0.2920	0.7827	54.65	0.3731	0.9519	45.65	0.6764	23.152	30.84
MoGe-2(L)	crop&fuse	196s	scale+shift	0.0835	0.3519	93.89	0.1134	0.2984	86.15	0.2141	12.644	72.04
DAV2(L)	crop&fuse	152s	scale+shift	0.1195	0.5704	86.34	0.1491	0.5633	78.90	0.2230	16.201	62.66
DA360(L)	end2end	0.026s	scale	0.0793	0.4407	94.07	0.0710	0.2930	93.53	0.1630	13.015	80.47

Appendix

Table S4: Comparison with crop&fuse baselines.

Figure S9: Additional visualizations on SUN360: six examples with depth maps and point clouds from our DA360.

Panel of Figure 21 — Figure S11: Visual comparison on example I of SUN360 ( Xiao et al. 2012 ) with SOTA depth estimation models.

Panel of Figure 22 — Figure S13: Visual comparison on example II of SUN360 ( Xiao et al. 2012 ) with SOTA depth estimation models.

Panel of Figure 23 — Figure S15: Visual comparison on example III of SUN360 ( Xiao et al. 2012 ) with SOTA depth estimation models.

Panel of Figure 24 — Figure S17: Visual comparison on example IV of SUN360 ( Xiao et al. 2012 ) with SOTA depth estimation models.

Panel of Figure 25 — Figure S19: Visual comparison on example V of SUN360 ( Xiao et al. 2012 ) with SOTA depth estimation models.

Panel of Figure 26 — Figure S21: Visual comparison on example VI of SUN360 ( Xiao et al. 2012 ) with SOTA depth estimation models.

Panel of Figure 27 — Figure S23: Visual comparison on example VII of SUN360 ( Xiao et al. 2012 ) with SOTA depth estimation models.

Panel of Figure 28 — Figure S25: Visual comparison on example VIII of SUN360 ( Xiao et al. 2012 ) with SOTA depth estimation models.

Panel of Figure 29 — Figure S27: Visual comparison on example IX of SUN360 ( Xiao et al. 2012 ) with SOTA depth estimation models.

Panel of Figure 30 — Figure S29: Visual comparison on example X of SUN360 ( Xiao et al. 2012 ) with SOTA depth estimation models.

Explore similar work

RePer-360: Releasing Perspective Priors for 360 $^\circ$ Depth Estimation via Self-Modulation
Mar 6, 2026Cheng Guan, Chunyu Lin, Zhijie Shen +2Monocular Depth Estimation Depth Estimation
Institute of Information Science, Beijing Jiaotong University, Beijing 100044, China. · Visual Intelligence +X International Cooperation Joint Laboratory of MOE, Beijing 100044, China.
DepthMaster: Unified Monocular Depth Estimation for Perspective and Panoramic Images
Jun 10, 2026Pengfei Wang, Shihao Wang, Liyi Chen +3Monocular Depth Estimation Omnidirectional Images
Visual Computing Lab The Hong Kong Polytechnic University
Unified Panoramic Geometry Estimation via Multi-View Foundation Models
May 25, 2026Vukasin Bozic, Isidora Slavkovic, Dominik Narnhofer +4Omnidirectional Images Robust Geometric Model Estimation
ETH Zürich · Google · Athlence Sports +1

Mar 6, 2026, cs.CVJ/K move · Enter open · S save

RePer-360: Releasing Perspective Priors for 360 $^\circ$ Depth Estimation via Self-Modulation

Cheng Guan, Chunyu Lin, Zhijie Shen, Junsong Zhang+1

Institute of Information Science, Beijing Jiaotong University, Beijing 100044, China. · Visual Intelligence +X International Cooperation Joint Laboratory of MOE, Beijing 100044, China.

Monocular Depth Estimation Depth Estimation 3D Perception Wireless Foundation Models Prior Knowledge Integration Modifications

Open paper AskPDF arXiv

Recent depth foundation models trained on perspective imagery achieve strong performance, yet generalize poorly to 360

^\circ

images due to the substantial geometric discrepancy between perspective and panoramic domains. Moreover, fully fine-tuning these models typically requires large amounts of panoramic data. To address this issue, we propose RePer-360, a distortion-aware self-modulation framework for monocular panoramic depth estimation that adapts depth foundation models while preserving powerful pretrained perspective priors. Specifically, we design a lightweight geometry-aligned guidance module to derive a modulation signal from two complementary projections (i.e., ERP and CP) and use it to guide the model toward the panoramic domain without overwriting its pretrained perspective knowledge. We further introduce a Self-Conditioned AdaLN-Zero mechanism that produces pixel-wise scaling factors to reduce the feature distribution gap between the perspective and panoramic domains. In addition, a cubemap-domain consistency loss further improves training stability and cross-projection alignment. By shifting the focus from complementary-projection fusion to panoramic domain adaptation under preserved pretrained perspective priors, RePer-360 surpasses standard fine-tuning methods while using only 1% of the training data. Under the same in-domain training setting, it further achieves an approximately 20% improvement in RMSE. The code is available at https://github.com/munimo/RePer360.

Abstract

Explore similar work

RePer-360: Releasing Perspective Priors for 360∘^\circ∘ Depth Estimation via Self-Modulation

DepthMaster: Unified Monocular Depth Estimation for Perspective and Panoramic Images

Unified Panoramic Geometry Estimation via Multi-View Foundation Models

RePer-360: Releasing Perspective Priors for 360 $^\circ$ Depth Estimation via Self-Modulation