Abstract
The target of this research is to experiment, iterate and recommend a systemthat is successful in recognition of American Sign Language (ASL). It is achallenging as well as an interesting problem that if solved will bring a leapin social and technological aspects alike. In this paper, we propose areal-time recognizer of ASL based on a mobile platform, so that it will havemore accessibility and provides an ease of use. The technique implemented isTransfer Learning of new data of Hand gestures for alphabets in ASL to bemodelled on various pre-trained high- end models and optimize the best model torun on a mobile platform considering the various limitations of the same duringoptimization. The data used consists of 27,455 images of 24 alphabets of ASL.The optimized model when ran over a memory-efficient mobile application,provides an accuracy of 95.03% of accurate recognition with an averagerecognition time of 2.42 seconds. This method ensures considerablediscrimination in accuracy and recognition time than the previous research.