fieldid E-Thesis & Research สถาบันเทคโนโลยีไทย-ญี่ปุ่น
สืบค้น:

เขตข้อมูล ข้อมูล
บทคัดย่อ
THAI SIGN LANGUAGE RECOGNITION FOR EMERGENCY REQUESTS USING DEEP LEARNING : Sign Language (SL) Sign Language (SL) is essential for communication within the Deaf and hard-of-hearing community. Thai Sign Language (TSL) possesses a unique structure, distinct from Chinese and American Sign Languages. It operates under its own grammatical rules, frequently utilizing simplified sentence structures for effective communication. Recent research in TSL recognitionhas explored various methodologies, including the interpretation of fingerspelling and the translation of sequential signs into text. This study aims to interpret continuous TSL into simple sentences to bridge the communication gap between Deaf and hearing individuals. Given the vast vocabulary of the Thai language and limited research time, this study introduces a TSL recognition system designed specifically for emergency scenarios, with data sampled from the most common emergency requests. Real-time sign language recognition (SLR) systems typically use computer vision to translate sequences of hand movements and body postures into text or speech instantaneously. The proposed system utilizes MediaPipe Holistic for hand tracking, integrated with Long Short-Term Memory (LSTM), Recurrent Neural Network (RNN), and Gated Recurrent Unit (GRU) architectures to facilitate robust sequence recognition. Developed using OpenCV, the system translates individual signs into text and displays them as simple sentences following Thai grammatical rules (verb + noun or noun + verb).
ผู้แต่ง
Nut Samaksaman
ประเภทสิ่งพิมพ์
Thesis (MIT) 2025
เลขหน้า
55
หัวเรื่อง
Information Technology--Thesis
หัวเรื่อง
Thai-Nichi Institute of Technology.Information Technology--Thesis
เอกสารฉบับเต็ม
Download full text