Sistem Deteksi Bahasa Isyarat Secara Realtime Dengan Tensorflow Object Detection dan Python Menggunakan Metode Convolutional Neural Network


  • Abraar Hayyu Gustsa S1 Teknik Informatika, Fakultas Teknologi Informasi, Universitas Hasyim Asy’ari
  • Ginanjar Setyo Permadi S1 Teknik Informatika, Fakultas Teknologi Informasi, Universitas Hasyim Asy’ari


Communicating using sign language is foreign to most people and very few people know how to communicate using sign language because it is not a mandatory language to learn. This case becomes a problem for people with special needs to interact and communicate with other people, especially people who do not know and understand how to communicate using sign language. Conventional Neural Network is the method that will be used in this study because this method is one of the deep learning methods that has the best current results in detecting an object, Deep Learning has excellent capabilities in computer vision. One of them is in the case of object classification in the image. By implementing one of the machine learning methods that can be used for object image classification, namely Convolutional Neural Network CNN. This is because the CNN method tries to imitate the image recognition system in the human visual cortex so that it has the ability to process image information. However, CNN, like other deep learning methods, has a weakness, namely the old model training process. In this study, the author uses a new dataset but is limited to only 4 sign language gestures (Hello, ILoveYou, Yes, No, Thanks). The results of testing the sign language training dataset reached an accuracy of 93.33%. In sign language predictions, the accuracy of the "Hello" symbol is 95%, "ILoveYou" is 93% accurate, "Yes" is 97%, "No" is 91% accurate, "Thanks" is 100% accurate.

Keywords: Sign Language, Convolutional Neural Network, Classification, Machine Learning, Deep Learning