Publication

Abstract : Efficient communication for the Deaf and hard-of-hearing community is primarily facilitated through sign language. An effective recognition system has the potential to empower these individuals by offering them seamless access to a myriad of services, education, and employment opportunities. However, a substantial gap exists in the literature, and there has been insufficient progress in developing practical and cost-effective Sign Language Recognition (SLR) models. We have proposed a novel method, the Intelligent Pattern Generation for Sign Language (IPGSL) algorithm, that utilizes pattern recognition principles to recognize sign language classes. The method generates a sequence of patterns that represent a sign language class. This article proposes an Identification of Key-frames (IKF) algorithm to detect significant key-frames from the input video of a sign language class, and subsequently generates patterns from the identified key-frames using the Intelligent Pattern Generation for Sign Language (IPGSL) algorithm. The proposed IPGSL model achieves an accuracy of 91.84% on the SMILE-DSGS dataset and 96.24% on the INCLUDE-50 dataset, demonstrating its effectiveness.