AI-Enabled Sign Language Detection: Bridging Communication Gaps for the Hearing Impaired

Nowadays people are so busy with their day-to-day work, so people cannot observe the needs and requirements of each other, which results in challenges for people like the deaf and dumb to talk with people. So now, to break the barrier between them, a solution for both gestures and signs mainly focuses on the detection of the sign or image caught on the web camera. The innovative project helps in determining the signs and has various uses which in turn helps people to grab the feelings, emotions, and thoughts of both deaf and dumb people. As a result, another informative tool emerges, offering significantly more efficient utilization. The whole project is about using many advanced features such as Convolution Neural Network (CNN) and Open- Source Computer Vision (Open CV). Each tool which is used plays an effective performance in the project. All the tools are widely accessed all over the world since it is user-friendly and also, they are multi- lingual support. The project undergoes different phases in order of sign detection i.e. when a sign is caught on the camera, the model detects the landmarks, and all the landmarks are connected and separated into frames later on image preprocessing is performed then the labels are detected from the hand, feature extraction helps in detecting labels, the whole process of image preprocessing and feature extraction are performed by CNN, Open CV. The final result of the model is the text of the particular hand gesture displayed. The proposed model gives an accuracy of highest 70.08% using pretrained model and by using CNN model gives an accuracy of 98.02%.