Indian Sign Language Interpretation for the Banking Sector Using CBAM-Enhanced YOLOv8

View at publisher Summarise with VIRAT
Authors: Karthikeyan G, Naveen Solomon J J, Iyshwarya Rathi K, Yogameena B
Cited by: 0
References: 14

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Particularly in service settings like banks, Indian Sign Language (ISL) is essential for facilitating communication between individuals with speech and hearing impairments. However, background clutter, uneven lighting, multiple signers, and motion variations continue to make real-time ISL recognition difficult. In order to improve spatial focus and feature extraction in complex scenes, this work presents a real-time ISL-to-text and speech translation system based on You Only Look Once Version 8 (YOLOv8) and enhanced with the Convolutional Block Attention Module (CBAM). To guarantee practical training, a custom dataset of 679 annotated photos covering seven banking gestures was gathered in actual banking environments. According to experimental results, the CBAM-integrated YOLOv8 outperforms the baseline YOLOv8 and current systems like MudraNet, achieving a mean average precision of 86.54% and a precision of 87.35%. The suggested model shows great promise for real-time deployment in banking environments to support accessible customer service due to its improved accuracy and low latency.

Fields of Study & Subjects

Hand Gesture Recognition Systems Advanced Neural Network Applications Human Pose and Action Recognition

Publication details

Journal / source
International Conference on Computing, Communication, Control and Cyber-Physical Systems (I5CPS)
Pages
1–6
Publication date
26 February 2026
Language
EN