Indian Sign Language Classification using Deep Convolutional Neural Network with Transfer Learning

The Indian Sign Language (ISL) classification model uses Convolutional Neural Networks (CNN) and analysis with Transfer Learning technology to understand ISL gestures by deaf people. This process entails categorizing hand signals into thirty-five groups, which comprise digits 1-9 (9) and alphabets A-Z (26). The developed model is a great support for people who want to know what the deaf say. The ultimate aim of this project is to categorize ISL Gestures into 35 distinct categories. These divisions are made using Convolutional Neural Networks (CNN). To achieve higher accuracy in classification, various transfer learning models such as ResNet50, Xception, DenseNet121, EfficientNetB5 and VGG19 were used in this study. Consequently, ResNet50 achieves an accuracy of 99.96%, Xception at 27.15%, DenseNet121 at 77.94%, and EfficientNetB5 at 50.93%.