Cost-effective Flex Sensor Glove-based Bangla Sign Language (BdSL) Interpreter for Resource-limited Settings

Abstract

In Bangladesh, approximately 13 million individuals experience hearing loss, including nearly 3 million with severe to profound impairments, creating significant communication barriers for the deaf and mute community. Most existing Bangla Sign Language (BdSL) interpretation systems rely on image-based approaches. However, these systems are sensitive to environmental conditions, require substantial computational resources, and often involve high implementation costs. To address these limitations, this paper presents a low-cost wearable BdSL interpreter based on flex sensor gloves and inertial sensing. The proposed system utilizes ten flex sensors to capture finger bending and two MPU6050 gyroscope modules to measure hand orientation. Sensor data are processed using a Threshold-Based Rule Classifier (TRC) algorithm running on an Arduino Mega 2560, eliminating the need for computationally intensive machine learning or image-processing techniques. The developed prototype recognizes a total of 25 BdSL gestures, comprising 10 numerals, 10 alphabet letters, and 5 commonly used words. The system provides both text and speech outputs to facilitate communication for hearing and visually impaired users. Experimental evaluation involving five volunteers demonstrated an overall recognition accuracy of 91.2%, with recognition accuracies of 98% for numerical gestures, 80% for alphabetical gestures, and 100% for word-level gestures. Confusion matrix analysis showed macro-F1 and weighted-F1 scores of 0.9798 for numerical gestures and 0.7991 for alphabetical gestures. The complete prototype was developed at a cost of approximately 100 USD using commercially available hardware components sourced from the local market. By avoiding dependence on cameras and high-performance computing platforms, the proposed system provides robust real-time performance under diverse environmental conditions while maintaining affordability and portability. The results demonstrate the potential of low-cost wearable technology to improve accessibility and promote inclusive communication for Bengali-speaking deaf communities in resource-limited settings.