Design and Analysis of Sign Language Gesture Recognition as Text: A Redesigned Artificial Neural Network Approach

Abstract

Numerous specialists tend to frame a gesture sign for words that may have discernible research result. In any case, because of alphabetic sign gesture prediction, numerous specialists neglects to close their exploration with huge result, this turns our research expectation towards an upgrade in compelling prediction over alphabetic sign gesture. In general, the researcher clear there research diagram in four stages information securing, preprocessing, feature extraction, and sign recognition. Moreover, with this consolidation of segmentation process before the feature, extraction will upgrade the viability of prediction accuracy definitely if there should be an occurrence of the ongoing procedure. The key procedure in this whole approach is sign recognition, where the specialist tend to use traditional systems in the midst of Artificial Neural Network (ANN) places an indispensable part. The proposed strategy raises the execution of the whole procedure in redesigning the genuine structure of ANN (hidden layer and neuron) by joining optimization techniques. This blend demonstrates a very clear outcome over customary procedures with its predicting accuracy and powerful execution. ANN in association with ABC reveals 96.71% for static data, while real time data set ANN associated with SSO reveal superior results of 83.73% over 75.75% by ANN-ABC.