Recognition of American Sign Language Using Deep Convolution Network

Around 466 million individuals are deaf or hard of hearing worldwide. Those individuals must engage with others, learn new things, and participate in activities. Sign language serves as a link between them and the rest of the world. There are numerous problems in creating a system that can automatically recognize sign language. For that, we have developed models for American Sign Language, we used a dataset that represents the American Sign Language Alphabet, which includes 29 classes for all English sign language letters as well as three additional characters: "del", "nothing" and "space." The collection contains 10,208 in image format, with each class presented in a separate folder. 80 percent of the dataset was used for training, while 20 percent was used for testing. The accuracy of the results for the proposed system was 100%.