شعبان ذكريايي عزيزي

This thesis focuses on developing an innovative and efficient model for sign language recognition using advanced deep learning technologies, aimed at creating a comprehensive platform to facilitate effective interaction and communication between deaf or speech-impaired individuals and society. Given the challenges in current sign language recognition systems, this research utilizes a combination of Convolutional Neural Networks (CNN) and the advanced YOLO model. By leveraging multi-level feature extraction capabilities and real-time image processing, it provides accurate detection of both static and dynamic sign language gestures. In this regard, the process of recognizing and translating complex hand and finger movements has been meticulously examined, and various deep learning algorithms have been eva‎luated to enhance the system's accuracy, speed, and adaptability. For the system's implementation, diverse datasets from American Sign Language (ASL) and Persian Sign Language (PSL) were collected and processed. Through data augmentation and transfer learning methods, the model was optimized to recognize hand movement features under varying environmental and lighting conditions. After thorough training and eva‎luation, the proposed model, leveraging advanced image processing and deep learning techniques, achieved a 95.3% accuracy rate in recognizing sign language gestures compared to other studies. The results demonstrate that this model not only offers high accuracy and speed but also exhibits strong stability and adaptability in challenging conditions relative to other references. Additionally, this research highlights the model's high potential for various applications, including Human-Computer Interaction (HCI) systems, real-time sign language translation tools, and assistive technologies for individuals with speech and hearing impairments. Thus, these achievements can play a pivotal role in enhancing the quality of life for disabled individuals and promoting communicative equality in societies.

زبان اشاره , يادگيري عميق , شبكه‌هاي كانولوشني عميق , تشخيص ژست دست

Sign Language Recognition , Deep Learning , Convolutional Networks , YOLO , Human-Computer Interaction

Shaban Zakariaei Azizi

Dr. Sattar Mirzakuchaki