"Hastabhashya"-Interpretation of Indian Sign Language using SqueezeNet and YoloV3

• Published in: 2023 International Conference on Emerging Trends in Networks and Computer Communications (ETNCC) pp. 1 - 6
• Main Authors: Anbazhagan, Adithya, V, Deeksha D, Kumar Reddy, Ambavaram Tharun, Singh, Paras Nath
• Format: Conference Proceeding
• Language: English
• Published: IEEE 16.08.2023
• Subjects: Assistive technologies; Auditory system; Convolutional Neural Networks; Deep learning; Face detection; Gesture recognition; Mobile applications; object detection; realtime performance; sign language recognition; SqueezeNet; Symbols; YOLOv3
• DOI: 10.1109/ETNCC59188.2023.10284968

The primary means of communication for persons who are hearing and voice-impaired is sign language, which is sadly not understood by a significant section of the public. To solve this problem, a naïve method is offered that converts sign language symbols into comparable plain text alphabets almost instantly using image processing and deep learning principles. Beyond providing healthcare for the deaf and dumb community, this technology has the potential to evolve into a general-purpose symbol-to-text converter with a variety of uses. The difficulties with character classification for Indian Sign Language (ISL) is tried to solve in this paper and work. Even though American Sign Language (ASL) has been the subject of much research, this cannot be said of Indian Sign Language (ISL) due to the absence of standardized datasets, obscured aspects, and regional variances in the language. ISL research has been hampered by these issues, but our initiative aims to fix them.