File size: 1,264 Bytes
5d5d3a5
 
da95b6d
0bac246
da95b6d
 
 
 
 
 
 
 
 
 
563d336
 
c619616
563d336
cc24617
563d336
 
 
 
 
 
 
 
 
 
 
 
 
 
bab4d44
563d336
 
 
 
 
bab4d44
 
563d336
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
---
license: apache-2.0

task_categories:
  - depth-estimation

tags:
  - depth-estimation
  - 3d-vision
  - multimodal
  - metric-depth

paper:
  - arxiv: 2605.15876
---

# DepthVLM-Bench

DepthVLM-Bench is a unified indoor-outdoor metric depth estimation benchmark designed for vision-language models (VLMs). The benchmark provides diverse indoor and outdoor scenes with metric depth annotations in a unified VLM-compatible format, enabling large multimodal models to jointly learn dense geometry prediction and multimodal understanding.

## Features

- Unified indoor and outdoor metric depth estimation
- VLM-compatible data format
- Dense depth supervision for multimodal foundation models
- Designed for scalable multimodal training

## Paper

[Unlocking Dense Metric Depth Estimation in VLMs](https://arxiv.org/abs/2605.15876)

## Usage

Please refer to the official repository for:

- Data preprocessing
- Evaluation scripts
- Visualization examples

Repository: https://github.com/hanxunyu/DepthVLM

## Citation

```bibtex id="83r6sk"
@article{yu2026unlocking,
  title={Unlocking Dense Metric Depth Estimation in VLMs},
  author={Hanxun Yu and Xuan Qu and Yuxin Wang and Jianke Zhu and Lei Ke},
  journal={arXiv preprint arXiv:2605.15876},
  year={2026}
}