Silent Conversations: Deep Learning for American Sign Language Recognition with VGG-16
Vijay Madaan, Neha Sharma, Rahul Chauhan, Hemant Singh Pokhariya
Abstract
This paper implements the deep learning technique VGG-16 convolutional neural network architecture that provides a method to correctly recognize ASL alphabet letters and symbols from images. The model successfully completes a long training procedure, achieving an accuracy of 84.97%, demonstrating its effectiveness in ASL gesture interpretation. To emphasize the model's durability and effectiveness in classification tasks, it has a relatively low loss value of 69.20%. This study provides a reliable method of ASL interpretation, which is a significant advancement in assistive technology for deaf and hard-of-hearing people. The findings indicate that deep learning technologies have the potential to improve accessibility and communication for those who engage primarily via sign language.