Integration of Deep Learning in Teaching Sign Language Russian Alphabet: Development of an Application for Children with Hearing and Speech Impairments

This thesis explores and develops effective methods for teaching dactylology and sign language for the Russian language. An analysis of existing solutions was conducted, revealing their shortcomings and proposing improvements using modern machine learning and deep learning algorithms. Data collection was carried out using a developed Telegram bot, ensuring high variability and quality of the data. Various algorithms, such as 2D-CNN, 1D-CNN, and traditional machine learning methods, were implemented and tested, resulting in the creation of a multifunctional interactive application based on Tkinter for dactylology training. The application includes modules for learning, recognition, testing, and exams, promoting effective and engaging learning. The conducted experiments confirmed the high efficiency of the proposed methods and algorithms. This work lays the foundation for further research and development in the field of sign language recognition and the creation of educational tools for deaf and hard-of-hearing individuals.