CENSUS-HWR: A Large Training Dataset for Offline Handwriting Recognition

Progress in Automated Handwriting Recognition has been hampered by the lack of large training datasets. Nearly all research uses a set of small datasets, which can cause models to overfit. We present CENSUS-HWR, a new dataset consisting of full English handwritten words in 1,812,014 grayscale images. A total of 1,865,134 handwritten words from a vocabulary of 10,711 unique words in the English language are present in this collection. This dataset is intended to serve handwriting models as a benchmark for deep learning algorithms. This huge English handwriting recognition dataset has been extracted from the US 1930 and 1940 censuses taken by approximately 70,000 enumerators each year. The dataset and the trained model with their weights are freely available to download at https://writeoff.cs.byu.edu/data.html.