Sign Language to Text Conversion using Hand Gesture Recognition

Most people are unaware of sign language which creates a language barrier amongst people and it's very challenging to find interpreters for American Sign Language (ASL). Therefore, this research is based in ASL fingerprint-based real-time method utilizing Convolutional Neural Networks (CNN). The approach discussed in this work involves passing the hand through a filter first, and then a classifier that determines the type of hand gesture once the filter has been applied. It is implemented using CNN as it is very effective at addressing computer vision issues. This proposed model achieved 99% accuracy on the first (MNIST) dataset and 96% on the second (ASL) dataset.