Recognition of text in natural images is important for a number of tasks, for example, augmented reality, self-driving cars, and adaptive systems. The main aim of this review is to determine the change of specification of text spotting due to new development in deep learning technology. The intention of this research is to give a brief overview of modern approaches to the text spotting issue and deep learning in particular, their development, and key aspects based on the analysis of the presented research. This review gathers the recently published work, compares and contrasts the various methods, and examines the collaborative and sole priorities, limitations, opportunities and future courses of the field. This work offers a clear comparison of many algorithms and offers solutions to the improvement of existing models, especially for low-frequency scripts like Gujarati, increasing the robustness against varying light conditions and complex backgrounds.