A Review of Methods for Estimating Depth Based on Various Application Scenarios

Siqi Huang

2025

Abstract

In computer vision, depth estimation is a crucial task. The task is to measure the distance between each pixel and the camera. The accuracy and efficiency of estimation tasks have undergone a significant improvement due to the rise of deep learning. There are many distinct application situations for depth estimation, and the task can be broadly classified into two categories: stereo image depth estimation and 2D image depth estimation, depending on the needs of each scenario. In this paper, HR-Depth, HybridDepth, SPIdepth in 2D image depth estimation methods and UniFuse, NLFB, PanoFormer, OmniFusion, HiMODE in stereo image depth estimation methods are introduced and analysed. In addition, NYU-Depth V2, KITTI in 2D image dataset and Stanford3D, and Matterport3D in stereo image dataset are introduced in detail, and the effectiveness of these two kinds of approaches is examined and contrasted using the widely used evaluation indices. It is found that HybridDepth and SPIdepth perform well in 2D image depth estimation, and NLFB and HiMODE perform better in stereo image depth estimation. Future research in depth estimation may focus more on depth estimation studies of stereo images, because stereo images usually contain richer depth information, which can allow for higher accuracy in depth estimation.

Paper Citation

in Harvard Style

Huang S. (2025). A Review of Methods for Estimating Depth Based on Various Application Scenarios. In Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE; ISBN 978-989-758-765-8, SciTePress, pages 204-211. DOI: 10.5220/0013681100004670

in Bibtex Style

@conference{icdse25,
author={Siqi Huang},
title={A Review of Methods for Estimating Depth Based on Various Application Scenarios},
booktitle={Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE},
year={2025},
pages={204-211},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013681100004670},
isbn={978-989-758-765-8},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE
TI - A Review of Methods for Estimating Depth Based on Various Application Scenarios
SN - 978-989-758-765-8
AU - Huang S.
PY - 2025
SP - 204
EP - 211
DO - 10.5220/0013681100004670
PB - SciTePress