Automatic Script Identification in the Wild
This addresses the problem of script identification in multilingual real-world scenarios for applications like translation and OCR, but it is incremental as it builds on existing deep learning techniques.
The paper tackles script identification at word or line levels in natural scenes by constructing a large-scale dataset with 10 languages and proposing a deep learning algorithm, which achieves superior performance compared to conventional methods like CNN and LLC.
With the rapid increase of transnational communication and cooperation, people frequently encounter multilingual scenarios in various situations. In this paper, we are concerned with a relatively new problem: script identification at word or line levels in natural scenes. A large-scale dataset with a great quantity of natural images and 10 types of widely used languages is constructed and released. In allusion to the challenges in script identification in real-world scenarios, a deep learning based algorithm is proposed. The experiments on the proposed dataset demonstrate that our algorithm achieves superior performance, compared with conventional image classification methods, such as the original CNN architecture and LLC.