| Dataset | Train | Validation | Test | Character-Level Annotation | Word-Level Annotation |
|---|---|---|---|---|---|
| Synthetic Word | 7,224,612 | 802,734 | 891,927 | No | Yes (Cropped Word) |
| SynthText in the Wild | 800,000 | No | No | Yes (Rectangle) | Yes (Quadrangle) |
Demo images of Synthetic word dataset.
This dataset consists of 9 million images covering 90k English words.
Demo images of SynthText in the wild dataset.
This is a synthetically generated dataset, in which word instances are placed in natural scene images, while taking into account the scene layout. The dataset consists of 800 thousand images with approximately 8 million synthetic word instances. Each text instance is annotated with its text-string, word-level and character-level bounding-boxes.