The recognition of Indian Sign Language (ISL) has developed at a fast rate owing to the development of deep learning and computer vision. Nevertheless, the majority of modern systems are interested in individual gesture recognition and do not have contextual modelling to support inference at the sentential level. Moreover, there is a lack of integration between the ISL recognition and the multilingual translation into the South Indian regional languages like Tamil, Telugu, Kannada and Malayalam. This paper is a systematic review of the research on the topic of ISL and sign language recognition published from 2020 to 2026, which combines the approaches of bibliometric analysis and qualitative synthesis. Based on PRISMA standards, the screening and analysis of relevant studies were conducted to find out the methodological trends, the limitations of the datasets, and the gaps in contextual and multilingual modelling. Findings suggest that convolutional neural networks and spatio-temporal models prevail, new transformer-based systems are being developed, there is a lack of large-scale annotated corpora of ISL, and no work has been done so far on integrating Dravidian languages. On the basis of these results, an architectural framework that allows integrating context-sensitive multimodal feature extraction, transformer-based sequence modelling, and regional language translation is suggested. The research paper is a contribution to inclusive artificial intelligence and strategic directions of scalable, multilingual systems of ISL in South India.
N et al. (Thu,) studied this question.