Update README.md
Browse files
README.md
CHANGED
|
@@ -27,6 +27,12 @@ The UNICE dataset is crucial for training a universal and generalized model for
|
|
| 27 |
- **Content**: High-quality sRGB images generated by fusing the MES using an ensemble of multi-exposure fusion (MEF) techniques.
|
| 28 |
- **Purpose**: Used as the target output (pseudo-GT) for supervised training of enhancement models.
|
| 29 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
## Sample Usage
|
| 31 |
|
| 32 |
To download the dataset using Git LFS:
|
|
|
|
| 27 |
- **Content**: High-quality sRGB images generated by fusing the MES using an ensemble of multi-exposure fusion (MEF) techniques.
|
| 28 |
- **Purpose**: Used as the target output (pseudo-GT) for supervised training of enhancement models.
|
| 29 |
|
| 30 |
+
## 3. `pseudoGT_arniqa.csv`
|
| 31 |
+
- **Type**: Pseudo Ground Truth Quality Scores
|
| 32 |
+
- **Content**: ARNIQA scores for each pseudoGT image, indicating perceptual quality.
|
| 33 |
+
- **Purpose**: Enables quality-aware selection of pseudoGTs. Low-quality samples (e.g., score < 0.5) can be filtered out to improve training.
|
| 34 |
+
|
| 35 |
+
|
| 36 |
## Sample Usage
|
| 37 |
|
| 38 |
To download the dataset using Git LFS:
|