Abstract:
Visually impaired people often face difficulty in navigation and daily tasks, especially when it comes to identifying surrounding objects or recognizing currency denominations. A smart detection app is designed for visually impaired people to navigate their surroundings safer and independently. This solution integrates object recognition and currency detection. The core objectives of this mobile application are to detect and recognize Pakistani currency denominations as well as to detect indoor objects and provide immediate voice feedback. The system architecture incorporates a camera module, lightweight deep learning models and a voice feedback system ensuring usability and accessibility for visually impaired users. Advanced deep learning techniques are used to ensure the safety and assistance of visually impaired people. For detection purposes, YOLOv8 model is used that generates a text output by labeling that particular object. For assistance of visually impaired users, text output is converted to voice feedback using Text-to-Speech library. The app ensures accurate and efficient identification of various denominations of indoor objects and currency. The design methodology follows an object-oriented, modular approach with UML modeling, ensuring maintainability and scalability. This application is developed with a focus on user friendliness and runs effectively on Android devices. Extensive testing demonstrates the reliability and potential of a system that significantly improves users’ daily lives by providing greater autonomy and reducing their dependence on others. A comprehensive literature review highlights related work in object detection and currency recognition, underscoring gaps in integration and localization. This app fills these gaps by offering an all-in-one solution requiring no external hardware. The combination of Pakistani currency detection and object detect detection makes
this mobile application extremely helpful for the visually impaired users
