Akademik Veri Yönetim Sistemi
2nd International Conference on Forthcoming Networks and Sustainability in the AIoT Era, FoNeS-AIoT 2024, İstanbul, Türkiye, 27 - 29 Ocak 2024, cilt.1035 LNNS, ss.1-10
Sign language is a physical language that enables people with disabilities to communicate with each other by using hand and facial movements as a whole to express themselves. It is very important that sign language is learned by everyone and used as a communication tool for the disabled to adapt to social life and to express themselves easily. For this reason, people’s learning of sign languages, which are specific to the country’s spoken language, will increase the quality of life of people with disabilities. In this study, 12981 images of the numbers 0–10 in Turkish Sign Language taken from different angles were used as a data set. In the last stage of the study, the detection of digits over images was carried out with CNN, Resnet-50, VGG-16, Densenet-201, and Inception-V3 deep learning architectures. In the study, an effective model of deep learning algorithms is proposed to determine which number an action corresponds to in sign language. Examining the models, VGG-16 and Densenet-201 were the architectures that gave the highest accuracy with 100% accuracy. After these architectures, Inception-V3 architecture comes with 99.91% success in determining the numbers. It has been seen that it is very successful in detecting numbers in Turkish Sign Language using deep learning models.