qcraftai / gedepth

GEDepth: Ground Embedding for Monocular Depth Estimation (ICCV 2023)

GEDepth: Ground Embedding for Monocular Depth Estimation

Xiaodong Yang, Zhuang Ma, Zhiyu Ji, Zhe Ren
GEDepth: Ground Embedding for Monocular Depth Estimation, ICCV 2023
[Paper] [Poster]

Get Started

Installation

Please refer to INSTALL for the detail.

Data Preparation

Please follow the instructions in DATA.

Training and Evaluation

Please follow the instructions in RUN.

Main Results

DepthFormer is used in this repo as the baseline to exemplify the improvement by the proposed GEDepth. Please refer to the paper for more results, in particular, on the generalization enhancement.

• KITTI

Model	Abs Rel	Sq Rel	RMSE	Checkpoint
Baseline	0.052	0.156	2.133	[Link]
GEDepth-Vanilla	0.049	0.144	2.061	[Google Drive] [Baidu Cloud]
GEDepth-Adaptive	0.048	0.142	2.044	[Google Drive] [Baidu Cloud]

• DDAD

Model	Abs Rel	Sq Rel	RMSE	Checkpoint
Baseline	0.152	2.230	11.051	[Link]
GEDepth-Vanilla	0.147	2.155	10.784	[Google Drive] [Baidu Cloud]
GEDepth-Adaptive	0.145	2.146	10.672	[Google Drive] [Baidu Cloud]

Citation

Please cite the following paper if this repo helps your research:

@inproceedings{yang2023gedepth,
  title={GEDepth: Ground Embedding for Monocular Depth Estimation},
  author={Yang, Xiaodong and Ma, Zhuang and Ji, Zhiyu and Ren, Zhe},
  booktitle={IEEE/CVF International Conference on Computer Vision (ICCV)},
  year={2023}
}

License

Copyright (C) 2023 QCraft. All rights reserved. Licensed under the CC BY-NC-SA 4.0 (Attribution-NonCommercial-ShareAlike 4.0 International). The code is released for academic research use only. For commercial use, please contact [e-mail].