Sign Language Recognition using unsupervised feature learning

Abstract: Sign Language Recognition is a game-changer for deaf-mute persons, and it’s been studied for years. Unfortunately, each study has its own set of restrictions and cannot be used commercially. Some studies have proven to be successful in identifying sign language, but commercialization is prohibitively expensive. Researchers are now paying more emphasis to building commercially viable Sign Language Recognition systems. Researchers conduct their studies in a variety of ways. It all begins with the data collection methods. Because of the high cost of a decent device, the data collecting method varies, but a low-cost method is required for the Sign Language Recognition System to be commercialised. The methodologies utilised to create Sign Language Recognition differ from one researcher to the next.

Keywords: RGB, Kinect, sign language, accuracy, algorithm, reasonable, dataset.