Eye Assistant : Using mobile application to help the visually impaired

Mobile devices or portable smart gadgets are largely used in our day-to-day life now. These devices can be very useful in helping people who are visually impaired and assist them to make their life easier. This paper presents an android application that recognizes objects using real time object and text detection by scanning them. The application can run on the device without any remote server. Our solution uses a voice feedback to tell the user about the detected object. The app doesn't need any photograph to detect the object. To enable robust recognition, we first segment the object from the background using tensorflow machine learning API based on iterative diagram cuts. We then formulate the recognition problem as an instance retrieval task and the user gets to know the object through a text to speech method. The system is designed to help a visually impaired person know about the environment around them. This app can be run on any low-end smartphone devices.