| pretty_name: UNICE Dataset | |
| tags: | |
| - image-enhancement | |
| - multi-exposure | |
| - low-light | |
| - exposure-correction | |
| - computer-vision | |
| - pseudo-ground-truth | |
| - contrast-enhancement | |
| license: cc-by-nc-4.0 | |
| task_categories: | |
| - image-enhancement | |
| language: | |
| - en | |
| dataset_info: | |
| features: | |
| - name: input_image | |
| dtype: image | |
| - name: pseudo_gt | |
| dtype: image | |
| splits: | |
| - name: train | |
| num_bytes: ~ | |
| num_examples: ~ | |
| size_in_bytes: ~ | |
| dataset_type: image | |
| annotations_creators: | |
| - machine-generated | |
| source_datasets: | |
| - original | |
| pretty_name: UNICE Multi-Exposure Dataset | |
| description: > | |
| The UNICE dataset is a large-scale dataset for universal image contrast enhancement. | |
| It contains multi-exposure sequences (MES) rendered from HDR raw images and their corresponding | |
| pseudo ground truths generated via multi-exposure fusion (MEF). The dataset supports training | |
| models for low-light enhancement, exposure correction, backlit enhancement, and LDR-to-HDR transformation. | |