Decoding Information: A Dual Modality Approach for Sign Language Recognition

Sign language is an important transmission method through which people who have hearing impairments or have difficulty in hearing properly communicate with other people. In sign language, each sign or gesture is assigned a different meaning. But communicating through sign language can be challenging as there are still crowds that have no idea about it therefore to solve this issue there are many sign language recognition tools present but they can be used by people who have a basic idea of sign language. A dual modality approach has been proposed, so this tool can be used by both the users who know sign language as they can use the sign-to-text recognition model which uses Convolutional Neural Network (CNN), and the users who do not know sign language can use the audio to text converter model which uses Speech Recognition Python Module. It can be used by a broader community. Our model has been trained on 26 alphabets and has achieved an accuracy of 98%.