Development of a Computational Vision Model trained with Convolutional Neural Networks for Sign Language Interpretation

Vargas R.; Fasabi F.; Ticona W. · Lecture Notes in Networks and Systems (2026) · DOI: 10.1007/978-3-032-22236-7_8

Resumen

Millions of people around the world are affected by hearing and speech disabilities, which create barriers to effective communication and participation in everyday life. According to global estimates, approximately 2.4 billion people will suffer from some degree of hearing loss by 2050. In Peru alone, more than 500,000 people live with communication disabilities, highlighting the urgent need for inclusive technological solutions. This research presents the development of a computer vision model capable of recognizing the Peruvian sign language by detecting hand gestures using convolutional neural networks (CNNs). The system captures visual information from a standard camera, processes the images through a preprocessing process, and classifies the gestures using CNN-based architecture specifically trained for this purpose. A customized dataset was created using locally recorded signs and enriched with publicly available data to improve generalization. The CNN achieved an accuracy of 94.86%, demonstrating a promising performance in recognizing a variety of hand gestures, which reinforces the potential of deep learning approaches to support accessible communication tools for people with hearing and speech disabilities. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Instituciones: Universidad ESAN; Universidad Tecnológica del Perú