Machine Learning in ASL Fingerspelling Recognition: A Literature Review

• Published in: IEEE International Symposium on Computational Intelligence and Informatics pp. 000055 - 000062
• Main Authors: Pinnington, Jamie, Souag, Amina, Hannan Bin Azhar, M A
• Format: Conference Proceeding
• Language: English
• Published: IEEE 19.11.2024
• Subjects: Adaptation models; American Sign Language; ASL fingerspelling recognition; Collaboration; Convolutional Neural Networks; dataset quality; Deep learning; Focusing; Hidden Markov models; machine learning; real-time recognition; Real-time systems; Recurrent Neural Networks; Sign language; sign language recognition; Systematic literature review; Training; Transformers
• ISSN: 2471-9269
• DOI: 10.1109/CINTI63048.2024.10830860

The recognition of American Sign Language (ASL) fingerspelling through machine learning has seen significant advancements over the past few years. This literature review explores various machine learning models and their applications in ASL fingerspelling recognition, focusing on the transition from hardware-based approaches to sophisticated deep learning models. Key models discussed include Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Transformers, highlighting their performance, challenges, and advancements. The review emphasises the importance of dataset quality and diversity, and the necessity for models to handle real-time, varied conditions for practical applications. The findings suggest that while significant progress has been made, challenges such as dataset variability, environmental conditions, and real-time processing remain. Future research should aim at developing more adaptive models and fostering collaboration between technology and the ASL community to enhance real-world communication for the deaf and hard of hearing.