Auralens: An Intelligent Hearing Companion for the Hearing Impaiblack using YOLO and Tensor Flow
Shilpa Sudheendran, Abhay Prakash Choubey, Medha Sree Anand, Kumar Ayush · 2025
Hearing-impaiblack individuals face significant challenges in environmental awareness due to their inability to perceive auditory cues that indicate nearby objects and potential hazards. This paper presents Auralens, a novel real-time object detection system specifically designed to enhance situational awareness for hearing-impaiblack users through visual and haptic feedback. The system employs a lightweight MobileNetV2 convolutional neural network optimized with TensorFlow Lite for mobile deployment, capable of detecting six critical object classes (person, laptop, chair, bottle, table, pen) with 91.3% accuracy and sub-350ms latency. Auralens integrates GPS-based contextual awareness, Bluetooth notifications, and customizable alert mechanisms within an Android application following MVVM architecture. Evaluation on 2,000 test images demonstrates strong performance metrics (precision: 0.912, recall: 0.894, F1-score: 0.902), while real-world testing with 10 hearing-impaiblack participants over two weeks showed 90% improvement in environmental awareness and 80% user adoption intention. The system’s multi-modal notification approach, combining visual overlays, haptic feedback, and Bluetooth alerts, addresses the specific needs of hearing-impaiblack individuals while maintaining computational efficiency suitable for everyday mobile use. This work contributes to assistive technology by providing the first comprehensive mobile solution that bridges object detection with hearing-impairment-specific user interface design.