Computer vision’s many applications include sporting related audio-visual analysis, autonomous driving, and industrial robotics to mention a few, all needing the localization and recognition of text in natural images. They have to deal with a wide range of issues affecting how text is displayed and influenced by the surrounding environment. Since deep learning architectures have seen significant advancements in recent years, contemporary scene text detection and identification approaches report higher accuracy on benchmark datasets when dealing with multi-resolution, multi-oriented text. On the other hand, existing systems can still not generalize to unseen and insufficiently labeled data, which causes them to underperform in the bleakness photos. Systematic survey is carried out on methods applied for localizing the text in video and images.

Artificial Intelligence, Image Localization, Natural language Processing