Translation Between Sign Language and Symbolic Recording

Nowadays, the progress in computer vision and natural language processing is likely to increase the interest in applications of automated sign language translation. This work focuses on two aspects of the translation: data gathering for creating a basic sign language concepts recognition system and selecting known parts or known synonymized parts from text for phrase translation. The data gathering is evaluated in terms of recognition quality in real-world scenario, which requires a tool for automated dataset collection and annotation. We have developed the possibility to give data to the tool remotely. We have formulated methods for text-to-sign language translation via sentence split and phrase translation using synonymic counterparts’ acquisition. In the end, we have formulated general pipeline for automated written language – sign language translation and provide a system for text-to-sign translation, which can be incorporated into an animation framework. The developed application has been compared with the most popular analogue.