EQUAL SPEAK (KEY IJP************012)

• Ms.s.jaya Shree Lakshmi ,Mrs.s.shyma,Ms.s.anugraha,Ms.s.abinaya

Abstract

Developed a recognition system for sign languages to bridge the communication gap among deaf and hearing-impaired people by putting into action transfer learning with EfficientNet, a high-performance CNN. This utilized large-scale pre-training knowledge, as seen in datasets such as ImageNet, to limit the training time required hence improving performance. Fine-tuning allowed gesture recognition optimally without compromising its powerful feature extraction. It implemented a Flask-based web application to support predictions on images and videos in real-time, while being viable within educational tools, communication, and social services. High dependability, the capability of being adaptable-the system achieved results that showed it is highly efficient, compared with previous works of traditional methods. The results confirm its contribution will have impacts for accessibility and communication inclusivity with room for scale into further data sets and application spaces in the future.