Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# UNICE Dataset Description
|
| 2 |
+
|
| 3 |
+
This is the dataset released with the paper:
|
| 4 |
+
|
| 5 |
+
**UNICE: Training A Universal Image Contrast Enhancer**
|
| 6 |
+
|
| 7 |
+
The dataset consists of two main components:
|
| 8 |
+
|
| 9 |
+
## 1. `UNICEdataset.zip`
|
| 10 |
+
- **Type**: Multi-Exposure Sequences (MES)
|
| 11 |
+
- **Content**: sRGB images rendered from HDR raw images using an emulated ISP pipeline.
|
| 12 |
+
- **Structure**: Each sequence contains multiple images of the same scene with varying exposure values (EVs), from -3EV to +3EV.
|
| 13 |
+
- **Purpose**: Serves as input data for training and evaluating exposure and contrast enhancement models.
|
| 14 |
+
|
| 15 |
+
## 2. `pseudoGT.zip`
|
| 16 |
+
- **Type**: Pseudo Ground Truths
|
| 17 |
+
- **Content**: High-quality sRGB images generated by fusing the MES using an ensemble of multi-exposure fusion (MEF) techniques.
|
| 18 |
+
- **Purpose**: Used as the target output (pseudo-GT) for supervised training of enhancement models.
|