Prototype App Mobile for Real Time American Sign Language Recognition Based on Deep Learning

The recognition of sign language alphabet through computer vision techniques has been widely studied. As a result of multiple studies, the CNN algorithm has demonstrated a very high learning rate. Other researchers have created and published a dataset consisting of approximately 200 images of static gestures of the American Sign Language alphabet. For this research work, the aforementioned dataset was used, and the gestures were captured and trained with CNN. Consequently, a significantly good recognition rate was achieved and migrated to a mobile application capable of recognizing the gestures. The ultimate goal is to recognize and transcribe words through a mobile device, which could be very useful in the practical teaching of the sign language alphabet, providing a significant breakthrough for a more complete learning of sign language.