ImmersiveDepth: A Hybrid Approach for Monocular Depth Estimation from 360 Images Using Tangent Projection and Multi-Model Integration

Sarshar Dorosti; Xiaosong Yang

doi:10.1109/vrw66409.2025.00342

ImmersiveDepth: A Hybrid Approach for Monocular Depth Estimation from 360 Images Using Tangent Projection and Multi-Model Integration

Sarshar Dorosti
, Xiaosong Yang

Belfast School of Art

Bournemouth University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

ImmersiveDepth is a hybrid framework designed to tackle challenges in Monocular Depth Estimation (MDE) from 360-degree images, specifically spherical distortions, occlusions, and texture inconsistencies. By integrating tangent image projection, a combination of convolutional neural networks (CNNs) and transformer models, and a novel multi-scale alignment process, ImmersiveDepth achieves seamless and precise depth predictions. Evaluations on diverse datasets show an average 37% reduction in RMSE compared to Depth Anything V2 and a 25% accuracy boost in low-light conditions over MiDaS v3.1. ImmersiveDepth thus establishes a robust solution for immersive technologies, autonomous systems, and 3D reconstruction.

Original language	English
Title of host publication	2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW)
Pages	1392-1393
Number of pages	2
DOIs	https://doi.org/10.1109/vrw66409.2025.00342
Publication status	Published (in print/issue) - 8 Mar 2025
Event	2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW) - The Palais du Grand Large, Saint-Malo, France Duration: 8 Mar 2025 → 12 Mar 2025 https://ieeevr.org/2025/

Conference

Conference	2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW)
Country/Territory	France
City	Saint-Malo
Period	8/03/25 → 12/03/25
Internet address	https://ieeevr.org/2025/

Bibliographical note

Publisher Copyright:
© 2025 IEEE.

Keywords

Monocular depth estimation
360-degree images
tangent projection
VR
AR
SfM
MVS

Access to Document

10.1109/vrw66409.2025.00342

Cite this

@inproceedings{96835faf7111471587172a1ba401d8cd,

title = "ImmersiveDepth: A Hybrid Approach for Monocular Depth Estimation from 360 Images Using Tangent Projection and Multi-Model Integration",

abstract = "ImmersiveDepth is a hybrid framework designed to tackle challenges in Monocular Depth Estimation (MDE) from 360-degree images, specifically spherical distortions, occlusions, and texture inconsistencies. By integrating tangent image projection, a combination of convolutional neural networks (CNNs) and transformer models, and a novel multi-scale alignment process, ImmersiveDepth achieves seamless and precise depth predictions. Evaluations on diverse datasets show an average 37\% reduction in RMSE compared to Depth Anything V2 and a 25\% accuracy boost in low-light conditions over MiDaS v3.1. ImmersiveDepth thus establishes a robust solution for immersive technologies, autonomous systems, and 3D reconstruction.",

keywords = "Monocular depth estimation, 360-degree images, tangent projection, VR, AR, SfM, MVS",

author = "Sarshar Dorosti and Xiaosong Yang",

note = "Publisher Copyright: {\textcopyright} 2025 IEEE.; 2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW) ; Conference date: 08-03-2025 Through 12-03-2025",

year = "2025",

month = mar,

day = "8",

doi = "10.1109/vrw66409.2025.00342",

language = "English",

isbn = "979-8-3315-1484-6",

pages = "1392--1393",

booktitle = "2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW)",

url = "https://ieeevr.org/2025/",

}

Dorosti, S & Yang, X 2025, ImmersiveDepth: A Hybrid Approach for Monocular Depth Estimation from 360 Images Using Tangent Projection and Multi-Model Integration. in 2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW). pp. 1392-1393, 2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW), Saint-Malo, France, 8/03/25. https://doi.org/10.1109/vrw66409.2025.00342

ImmersiveDepth: A Hybrid Approach for Monocular Depth Estimation from 360 Images Using Tangent Projection and Multi-Model Integration. / Dorosti, Sarshar; Yang, Xiaosong.
2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW). 2025. p. 1392-1393.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - ImmersiveDepth: A Hybrid Approach for Monocular Depth Estimation from 360 Images Using Tangent Projection and Multi-Model Integration

AU - Dorosti, Sarshar

AU - Yang, Xiaosong

PY - 2025/3/8

Y1 - 2025/3/8

N2 - ImmersiveDepth is a hybrid framework designed to tackle challenges in Monocular Depth Estimation (MDE) from 360-degree images, specifically spherical distortions, occlusions, and texture inconsistencies. By integrating tangent image projection, a combination of convolutional neural networks (CNNs) and transformer models, and a novel multi-scale alignment process, ImmersiveDepth achieves seamless and precise depth predictions. Evaluations on diverse datasets show an average 37% reduction in RMSE compared to Depth Anything V2 and a 25% accuracy boost in low-light conditions over MiDaS v3.1. ImmersiveDepth thus establishes a robust solution for immersive technologies, autonomous systems, and 3D reconstruction.

AB - ImmersiveDepth is a hybrid framework designed to tackle challenges in Monocular Depth Estimation (MDE) from 360-degree images, specifically spherical distortions, occlusions, and texture inconsistencies. By integrating tangent image projection, a combination of convolutional neural networks (CNNs) and transformer models, and a novel multi-scale alignment process, ImmersiveDepth achieves seamless and precise depth predictions. Evaluations on diverse datasets show an average 37% reduction in RMSE compared to Depth Anything V2 and a 25% accuracy boost in low-light conditions over MiDaS v3.1. ImmersiveDepth thus establishes a robust solution for immersive technologies, autonomous systems, and 3D reconstruction.

KW - Monocular depth estimation

KW - 360-degree images

KW - tangent projection

KW - VR

KW - AR

KW - SfM

KW - MVS

UR - https://www.scopus.com/pages/publications/105005153759

U2 - 10.1109/vrw66409.2025.00342

DO - 10.1109/vrw66409.2025.00342

M3 - Conference contribution

SN - 979-8-3315-1484-6

SP - 1392

EP - 1393

BT - 2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW)

T2 - 2025 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW)

Y2 - 8 March 2025 through 12 March 2025

ER -

ImmersiveDepth: A Hybrid Approach for Monocular Depth Estimation from 360 Images Using Tangent Projection and Multi-Model Integration

Abstract

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this