VictorG-028/Small_subset_of_Synthetic_Word_Dataset

This dataset contains:

• 40351 images (71.39%) in train dataset
• 6378 images (14.29%) in validation dataset
• 6391 images (14.32%) in test dataset
• Total: 53120 images
• sourced from the extensive Synthetic Word Dataset, a large-scale word-image dataset.

The original and complete dataset (9 million images, 10.68GB) can be found and downloaded at this academic torrent.