A Mobile Application for Printed New Tai Le Character Recognition Based on Lightweight ResNet

With the development of deep learning, there are many mature applications in the field of image recognition, including the aspect of character recognition. The research on the recognition of ethnic minority characters is not only beneficial for daily character information processing, but also of great significance for protection and inheritance of cultural. However, there is almost no research on the recognition of the new Tai Le characters in Dehong. Therefore, this paper mainly uses OpenCV to build a printed character database of the New Tai Le character, which contains 40,000 character images. Then, the Resnet18 model is adopted to complete character recognition and achieves a 99.5% accuracy rate. Finally, we have developed two mobile applications, including an offline version and an online version based on the B/S architecture.