EMNIST Handwritten Digits Dataset

EMNIST stands for Extended MNIST, which is an extended dataset of NIST Special Database 19. This dataset converts images into 28×28 pixel image format and a dataset structure that matches the MNIST dataset.

The data volume of this dataset is 4 times that of MNIST, including:

• EMNIST ByClass: 814,255 characters, 62 unbalanced classes.
• EMNIST ByMerge: 814255 characters, 47 imbalanced classes.
• EMNIST Balanced: 131,600 characters, 47 balanced classes.
• EMNIST Letters: 145,600 characters, 26 balanced classes.
• EMNIST Digits: 280,000 characters, 10 balanced classes.
• EMNIST MNIST: 70,000 characters, 10 balanced classes.