Aplikasi Penerjemah Bahasa Isyarat Indonesia menjadi Suara berbasis Android menggunakan Tensorflow

Abstract

Indonesian Sign Language or BISINDO is a two-handed sign language that is used as a liaison in communication. BISINDO is used by people who have limited speech or hearing, but not for other communities. This causes BISINDO users have difficulties in conveying information because only a few people understand BISINDO. Therefore, an application was developed to help communication between BISINDO users and Indonesian in realtime. BISINDO classification is carried out using the Convolutional Neural Network method and the MobilenetV2 architecture using tensorflow. The classification results are used as a model for android which is then used as a sound. Based on model testing, the resulting accuracy rate resulted in 54.8% in the classification of 30 specified languages. Thus, the performance of the model can be said to be not optimal in classifying. Based on the application testers to 30% of respondents, it was found that respondents strongly agreed with this application with an average value of 83.95%.