Real-Time Sri Lankan Static Sign Language System using EfficientNet-B0

Sign Language serves as an important medium of communication for the deaf and hard of hearing community. However, there is a lack of understanding among the public regarding the various signs used. Although around 300,000 Sinhala-speaking individuals are either deaf or hard of hearing, dedicated applications for Sri Lankan Sign Language detection are quite limited. In contrast, various sign language detection systems for other popular sign languages, such as American Sign Language (ASL) and British Sign Language (BSL), exist in abundance. This study aims to design and develop a lightweight mobile application, backed by the Efficient-B0 model, for Sri Lankan Static Sign Language detection. This application can detect five popular words using nine static signs and can be scaled up in the future. The developed application was found to be lightweight, efficient, and usable among the target users, achieving an accuracy of over 90% for real-time sign detection.