Connecting Sign Language and Speech with Multilingual Translation Using Random Forest
Selva Sheela K, R. Praveena, H Sivaramana, M Rujuta, Vaishnavi R C · INTERANTIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING AND MANAGEMENT · 2024
This project demonstrates an innovative web application that aims to deepen the communication between gesture based clients and users of communicating through language. The application relies on advanced PC vision and AI algorithms for a seamless interpretation between gesture and voice communications. By including MediaPipe for hand gesture recognition and the Irregular Backwoods computation for movement classification, the system can understand and translate motions into equivalent words in Tamil and English. The app makes use of a webcam that captures hand movements, feeds them through a pre-trained Irregular Backwoods model, and then produces the interpreted gesture-based communication in text form. The constant nature of the application allows clients to share easily and multilingual support further fosters openness. The model's ability to be physically prepared and changed ensures versatility and accuracy in a vast range of settings. With this innovation, the drive aims to fill communication gaps and increase accessibility for the deaf and hard of hearing community, along with overcoming the limitations of the current sign language interpretation systems. Keywords: Real-time Sign Language Translation 1, Machine Learning 2, Random Forest 3, MediaPipe 4, Multilingual Support 5.