Portmanteauing Features for Scene Text Recognition

Scene text images have different shapes and are subjected to various distortions, e.g. perspective distortions. To handle these challenges, the state-of-the-art methods rely on a rectification network, which is connected to the text recognition network. They form a linear pipeline which uses text rectification on all input images, even for images that can be recognized without it. Undoubtedly, the rectification network improves the overall text recognition performance. However, in some cases, the rectification network generates unnecessary distortions on images, resulting in incorrect predictions in images that would have otherwise been correct without it. In order to alleviate the unnecessary distortions, the portmanteauing of features is proposed. The portmanteau feature, inspired by the portmanteau word, is a feature containing information from both the original text image and the rectified image. To generate the portmanteau feature, a non-linear input pipeline with a block matrix initialization is presented. In this work, the transformer is chosen as the recognition network due to its utilization of attention and inherent parallelism, which can effectively handle the portmanteau feature. The proposed method is examined on 6 benchmarks and compared with 13 state-of-the-art methods. The experimental results show that the proposed method outperforms the state-of-the-art methods on various of the benchmarks.

I. Introduction

Scene text recognition (STR) is the task of reading text from natural scenes. The semantic property of text in images embodies additional information that is useful for many practical applications such as intelligent document processing, street signs reading, and product recognition [1]. Because of these important applications, the field has drawn a great amount of attention from researchers and practitioners resulting in the emergence of various robust reading competitions [2]–[6]. STR is challenging because the images are acquired in a generally unstructured and uncontrolled setting. This, in turn, introduces more complexity and variability than traditional optical character recognition (OCR) tasks [1], [7], [8].