|
|
--- |
|
|
license: cc-by-nc-4.0 |
|
|
language: |
|
|
- en |
|
|
base_model: |
|
|
- depth-anything/Depth-Anything-V2-Small |
|
|
tags: |
|
|
- depth |
|
|
- relative depth |
|
|
- depth anything |
|
|
pipeline_tag: depth-estimation |
|
|
--- |
|
|
|
|
|
# DepthAnything-AC |
|
|
|
|
|
## Introduction |
|
|
[DepthAnything-AC](https://arxiv.org/abs/2507.01634) is a robust monocular depth estimation model fine-tuned from Depth-Anything-V2-Small, designed to enable zero-shot relative depth estimation under all-weather conditions. |
|
|
|
|
|
## Installation |
|
|
``` |
|
|
git clone https://github.com/HVision-NKU/DepthAnythingAC.git |
|
|
cd DepthAnythingAC |
|
|
conda create -n depth_anything_ac python=3.9 |
|
|
conda activate depth_anything_ac |
|
|
pip install -r requirements.txt |
|
|
``` |
|
|
## Usage |
|
|
You may refer to our [github](https://github.com/HVision-NKU/DepthAnythingAC) repo for detailed inference scripts. |
|
|
|
|
|
## Citation |
|
|
If you find this work useful, please consider citing: |
|
|
``` |
|
|
@article{sun2025depth, |
|
|
title={Depth Anything at Any Condition}, |
|
|
author={Sun, Boyuan and Modi Jin and Bowen Yin and Hou, Qibin}, |
|
|
journal={arXiv preprint arXiv:2507.01634}, |
|
|
year={2025} |
|
|
} |
|
|
``` |
