<![CDATA[Technological developments in the era of artificial intelligence have led to the development of computer systems capable of identifying sign language. Sign language is the primary means of communication for deaf and hard of hearing people used by millions of people around the world. This research aims to improve accuracy in alphabetic sign language recognition by using the Deep-CNN model. The method in this research starts from the selection of sign language datasets based on previous research. The dataset used in this study comes from Kaggle regarding American Sign Language which contains each training and test class representing a label (0-25) This dataset contains 27,455 training data and 7,172 test data. This research uses the Python programming language in performing data splitting, scaling, data augmentation, training, and evaluating. The architectural model built in this research is the Deep CNN architecture which is implemented to carry out the process of improving the accuracy of sign language recognition classification. The test results show an increase in the accuracy of alphabetic sign language classification compared to previous research. The increase in accuracy in the value of the Deep CNN model built managed to reach an accuracy rate of 99.72%. The model that has been built is the best model among previous research models.]]>
Copyrights © 2024
