20 November 2025 in Franklin Open
1 January 2022 in SSRN Electronic Journal
The primary source of communication used by hard-of-hearing people is sign language. The task of automatically recognising sign language motions is difficult. For the recognition of Indian sign language, a model based on hand anatomy and neural networks is proposed in this research. Features are extracted through FAST and SIFT techniques, and hand anatomy is utilised for the selection of hand landmarks out of these detected points. Finally, training and testing of the model have been done by the neural network. Experimentation was performed on challenging and publicly available datasets for isolated single-letter gestures (alphabets and numbers) and word gestures (various real-life gestures) with different backgrounds and lighting conditions. Experimental results prove that fast and highly accurate results are achieved by the proposed model when compared with other available models. An accuracy of 98.90% has been achieved for isolated gestures and 97.58% for word gestures.
20 November 2025 in Franklin Open