Drishti

Android, Node, Machine Learning

About

Project Drishti is an Android based application which is developed for Visually Impaired people, to assist them with various activities. Visually Impaired people face numerous hurdles in their day to day functioning. With this app we address the following enumerated hurdles of - 1) Outdoor navigation 2) Face Recognition 3) Currency Denomination Detection.

Drishti App:

Features

Outdoor Navigation :- Geolocation Technology is used to mark the destination for the user and navigate the user through the shortest available path. A 3D audio assistance provides continuous response to the user about the direction on which user has to walk. The 3D audio will be rendered dynamically to keep user updated about the direction.
Facial Recognition : Hybrid approach is used for implementing the face recognition system in Drishti. This approach is about using face descriptor method and probability-based similarity method for face recognition. In this approach, input to the system will be max 2-3 images of the person and on the basis of this, training of the images is be done and after that can be used to recognize the targeted face.
Currency Denomination Detection : Feature Extraction based approach is be used for currency denomination detection. In this approach nearly 43 features are extracted from a currency note and will be checked with features of that note in the database, thus proper output will be displayed.
Speech based Functionality :- All the above mentioned features can be utilized by speech, the user isn't required to touch the screen to find and use any feature. He/She can speak what is required and the app will perform the specified task.