Sign Language Recognition System using Convolutional Neural Network

Sign language is the common communication language for the hearing and speech-impaired community. It is hard for most people to communicate in sign language without an interpreter. Sign language refers to the tracking and identification of meaningful human expressions made with the hands, arms, fingers, heads, etc. The method used in this case converts the sign language movements into a spoken language that the listener may easily understand. The communication using sign language is useful for the peoples depend on gestural sign language but it is more complex for the other publics. The existing systems are not efficient since they are struggling with skin tone detection. But, adding a filter symbol can be recognized regardless of skin tone. In this work, primarily focused on analyzing convolutional neural networks (CNN). There are four kinds of layers: convolution layers, fully connected layers, pooling/subsampling layers and nonlinear layers for learning new characteristics.