Speech Recognition based detection of Real-time Objects with Tensor Flow

Authors

  • Vijay Karnatak, Satwik Teotia, Renu Yadav

Keywords:

Blind, Deep Learning, Object Detection, Object Classification, Convolutional Neural Network

Abstract

A system for persons who are blind or visually impaired that runs on Android. For users who are visually impaired, it offers object detection in the close vicinity. This technique aids those who are blind in recognising everyday objects such as a chair, table, phone, etc. in their environment. This system helps the blind person buy at the supermarket by using an RGB colour camera to capture images of the immediate area and deep learning to recognise the type and placement of objects in front of the blind person. This method builds two sets of picture databases with a combined total of nine categories, each with an object detection system based on an SSD network as well as an object classification system based on a MobileNets network, to train a neural network. In order to install an object categorization network on an Android phone, this system is constructed of an Android - TensorFlow interface. The Android terminal now has a voice announcement feature that notifies the blind person in real time when an object is recognised. For persons with vision impairments, this system has been shown to be more effective.

Published

2023-06-25

How to Cite

Vijay Karnatak, Satwik Teotia, Renu Yadav. (2023). Speech Recognition based detection of Real-time Objects with Tensor Flow. SJIS-P, 35(3), 136–143. Retrieved from http://sjisscandinavian-iris.com/index.php/sjis/article/view/650

Issue

Section

Articles