Machine Learning Models Applied in Sign Language Recognition

One of the most relevant worldwide problems is the inclusion of people with disabilities. In this research we want to help focusing in the people with hearing disabilities, being able to translate sign language into words that we could read. It is a common worldwide problem to be able to accurately predict the gestures of non-hearing people in order to be able to communicate efficiently with them and not have a barrier when they want to perform their daily activities. In order to that we propose a three phase method combining Data preparation(The dataset used for this is the "Australian Sign Language sings", which is public and free to use) and cleaning phase, modeling using Random Forest Vector Support Machine and Neural Networks, able to optimize and qualify these models using the measures of accuracy, precision, recall and f1-score. Therefore, in this work we try to offer the highest possible quality measures to the prediction of signs in the Australian language with the mentioned dataset. This also opens the way for future research where more advanced supervised modeling techniques can be applied to improve the values obtained.