Abstract : Our proposal accentuates to help blind people to navigate independently using real-time indoor object detection and identification model. The project aims to help blind and visually impaired people to detect and recognize objects around them, through a small camera that is fixed on the glasses. The device setup gives an audio output that is sent to an earphone placed on the person’s ear to help him/her find various items easily and independently. All the captured data will be fed to raspberry pi that will have an inbuilt code for object detection in python language and will contain Opencv and TensorFlow files in it. The unit will analyze the image based on the ssd_mobilenet network and custom data-set and the processed data that is the name of the object will come as output. This output will be heard by the user in voice format by incorporating TTS in our model. This model will act as an aid for blind people and will help them to do the daily chores easily and independently.

Keywords- Tensorflow, Text to Speech converter, Raspberry pi, SSD_ Mobilenet.


PDF | DOI: 10.17148/IARJSET.2021.86106

Open chat