Sign Language Recognition using Machine Learning
1Sameer Metkar, 2Shubham More, 3Durgesh Kolhe, 4Omkar Mandavkar,5Prof. Aarti Abhyankar
1Student, K.C. College of Engineering and Management Studies and Research, University of Mumbai, Thane
2Student, K.C. College of Engineering and Management Studies and Research, University of Mumbai, Thane
3Student, K.C. College of Engineering and Management Studies and Research, University of Mumbai, Thane
4Student, K.C. College of Engineering and Management Studies and Research, University of Mumbai, Thane
5Guide, K.C. College of Engineering and Management Studies and Research, University of Mumbai, Thane
Abstract:. Communication is a very important aspect of society. It is said that Man is a Social animal, and for this to be literal an efficient communication is very necessary. Similarly in the stone age, communication happened using drawings etc and as we progressed , society developed script, languages which eventually led to the growth of society. With time globalization started and the spread of languages began, The languages then were classified into verbal and non verbal. People who travelled and couldn’t understand a language used non verbal forms to communicate in foreign lands. But Non Verbal languages are not only meant for that but also the only method for People With Hearing and Speaking Disabilities (here onwards mentioned as PWHSD) to communicate without external tools. And since they are small percent of the consensus they are often disregarded or not thought of generally. Though they are a small percent but definitely not a negligible share. To have conversations with them keeping in mind that not everyone can learn another new non verbal language for communicating with them, to facilitate virtual communication with them we have developed a model that can help PWHSD individuals to communicate with others and also help others understand what they are saying. For this we have used a Mediapipe model that can in real time recognize sign language. With the amount of dataset we have provided the accuracy of the model is pretty good.
Index terms: Sign Language Recognition (SLR), Computer Vision, Machine Learning, American Sign Language, TensorFlow, Mediapipe, LearnSl.