Synthetic Data and Artificial Neural Networks for Natural Scene Text Recognition

TLDR

The paper proposes a framework for recognizing natural scene text. The framework trains deep neural networks solely on realistic synthetic data, using holistic image input and three word‑reading models (dictionary, character sequence, bag‑of‑N‑grams) without human labels. The approach achieves state‑of‑the‑art results on language‑based and unconstrained text datasets, outperforming prior methods while incurring no data‑acquisition cost.

Abstract

In this work we present a framework for the recognition of natural scene text. Our framework does not require any human-labelled data, and performs word recognition on the whole image holistically, departing from the character based recognition systems of the past. The deep neural network models at the centre of this framework are trained solely on data produced by a synthetic text generation engine -- synthetic data that is highly realistic and sufficient to replace real data, giving us infinite amounts of training data. This excess of data exposes new possibilities for word recognition models, and here we consider three models, each one reading words in a different way: via 90k-way dictionary encoding, character sequence encoding, and bag-of-N-grams encoding. In the scenarios of language based and completely unconstrained text recognition we greatly improve upon state-of-the-art performance on standard datasets, using our fast, simple machinery and requiring zero data-acquisition costs.