| dataset_info: | |
| features: | |
| - name: image | |
| dtype: image | |
| - name: question | |
| dtype: string | |
| - name: answer | |
| dtype: string | |
| splits: | |
| - name: train | |
| num_bytes: 6328540620.853265 | |
| num_examples: 379594 | |
| - name: validation | |
| num_bytes: 1582143491.1467357 | |
| num_examples: 94899 | |
| download_size: 7864843580 | |
| dataset_size: 7910684112.0 | |
| configs: | |
| - config_name: default | |
| data_files: | |
| - split: train | |
| path: data/train-* | |
| - split: validation | |
| path: data/validation-* | |
| task_categories: | |
| - image-to-text | |
| - visual-question-answering | |
| language: | |
| - dv | |
| tags: | |
| - thaana | |
| - dhivehi | |
| - ocr | |
| pretty_name: ocr_thaa | |
| license: apache-2.0 | |
| # Dhivehi Single-Line Text-Image Dataset | |
| A collection of synthetic Dhivehi text images for training and evaluating text-image / vision models etc. | |
| Each image contains a single line of Dhivehi text with various visual styles and augmentations (Check the config field for more info on the row). | |
| ***Note: This dataset is a subset from alakxender/dhivehi-vrd-images.*** | |