Design and Implementation of Visually Impaired Assistant system

The need for independent living is recognised in the case of visually impaired people who face a variety of issues, including social isolation.They are forced to live in strange environments with no means of assistance. Today’s tech world should supply them with this assistance for easier socialization. Our aim is to facilitate the life of visually impaired people who could lack the technology to help them in their lives. This project is proposed to focus on the development of a mobile application that uses voice commands and text to voice technology to enable users to interact with the mobile application. The user captures the item they want to identify using the smartphone’s camera and pick the image they want to upload for processing. The image chosen then will undergo through pre-processing techniques depending on the feature the user chose, the features include Money recognition CNN MobileNet architecture was used to analyze the Egyptian banknotes and Color and patterns recognition CNN resnet50 architecture was used to analyze the color and pattern of the clothing. As well, the application allows the user to register or login using facial recognition to further facilitate the experience for users. The application is designed to be effective and can be easily navigated by visually impaired users. ResNet50 architecture produce the highest accuracy in patterns and colors cloth recognition, Mobilenet architecture produce the highest accuracy in currency recognition.