Comprehensive Study on Recent Advances and Future Trends in Detection & Recognition of Scene Text

Abstract

Increasing use of smart phone in our daily life to capture images initiates a need to recognize text from natural images which is nowadays a promising research topic in the field of computer vision and document analysis due to its various applications. Text in natural scenes exists in almost every phase of our daily life. From the front of the buildings in our locality to the cover of a book in library. Certainly, text is one of the most radiant and powerful creations of human kind. Despite of variety of challenges and issues, especially in current era, lot of advancement has been taken place in terms of techniques and datasets use for detection and recognition of text.

The purpose of this survey is (1) introduce recent works, (2) identify modern algorithms and datasets, (3) identify different adversarial attacks on text recognition, (4) identify the different libraries supporting text detection and recognition task, (5) next generation applications and predict research guidelines for the future. Additionally, this manuscript offer in detail comparative analysis of recently used techniques and benchmark datasets. This study can offer a noble orientation for beginner researchers in the field of detection and recognition of scene text.