A Survey of ASL Interpretation Optimizer

Abstract

Since most individuals do not know sign language and interpreters are very hard to find, we have developed a real-time system for American sign language that is based on fingerspelling. Sign language is one of the oldest and most natural forms of language for communication. In our approach, the hand is first put through a filter and then put through a classifier, which determines the type of hand motions. We are attempting to translate sign language using Python based on CNN in order to reduce the verbal communication gap between D&M and non-D&M people and to ensure effective communication among all. Additionally, it gives deaf persons the chance to communicate verbally with vocal people without the use of an interpreter. The system is designed to automatically translate ASL. We trained the model in three optimizers (Adam optimizer, SGD optimizer and Adadelta optimizer) and compared the accuracy obtained from all these optimizers. The interpretation of British, Indian, and American sign languages has been the subject of numerous previous efforts. However, we will approach this project differently and use a novel classification method that increases accuracy.