NextGenConnect: Indian Sign Language Using 5-Layer CNN and MobileNetV2 Transfer Learning

For individuals with hearing or speech disabilities sign language is an important form of communication tool, yet technological solutions for interpreting these gestures remain limited in both scalability and accuracy. This paper introduces NextGenConnect, a deep learning–based framework that recognizes Indian Sign Language (ISL) gestures using a two-stage architecture. The first step involves a custom-built 5-layer Convolutional Neural Network (CNN) used for feature extraction. Then, transfer learning is applied using MobileNetV2 to refine and classify gestures efficiently. A validation accuracy of 98.86% was achieved by our system and is optimized for real-world deployment, even on resource-constrained devices. The results demonstrate the feasibility of building compact and accurate gesture recognition systems for assistive communication. Potential use cases include accessibility tools for education, virtual interpreters, communication interfaces in public services, and real-time assistance for individuals facing impairments in hearing or speech.