Update README.md
Browse files
README.md
CHANGED
|
@@ -30,12 +30,6 @@ We present Depth Anything at Any Condition (DepthAnything-AC), a foundation mono
|
|
| 30 |
|
| 31 |
To address the lack of high-quality annotations in corrupted scenes, we introduce a lightweight **unsupervised consistency regularization** framework that enables training on unlabeled data. Additionally, our proposed **Spatial Distance Constraint** helps the model learn patch-level geometric relationships, enhancing semantic boundaries and fine details.
|
| 32 |
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
<div align="center">
|
| 36 |
-
<img src="https://github.com/HVision-NKU/DepthAnythingAC/blob/main/assets/depthanything-AC-video.gif?raw=true" alt="video" width="100%">
|
| 37 |
-
</div>
|
| 38 |
-
|
| 39 |
|
| 40 |
## Installation
|
| 41 |
|
|
|
|
| 30 |
|
| 31 |
To address the lack of high-quality annotations in corrupted scenes, we introduce a lightweight **unsupervised consistency regularization** framework that enables training on unlabeled data. Additionally, our proposed **Spatial Distance Constraint** helps the model learn patch-level geometric relationships, enhancing semantic boundaries and fine details.
|
| 32 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
|
| 34 |
## Installation
|
| 35 |
|