Asteriks/chars74k-eng-good

Chars74k

The "Good" subset of the "English" subset of the Chars74k dataset split into training and validation sets.

The validation set was created to match the label distribution of the training set.

62 classes (0-9, A-Z, a-z)

Dataset page: https://teodecampos.github.io/chars74k/

Paper describing the dataset:
https://www.semanticscholar.org/paper/Character-Recognition-in-Natural-Images-Campos-Babu/dbbd5fdc09349bbfdee7aa7365a9d37716852b32

5 images where removed due to poor quality.

Lable distribution: