Skip to content

Object Detection for People with Visual Impairment

Team Members

  • Shuohong Liu
  • Yichen Zhu

Abstract

We built a helmet that can guide visually impaired people to walk freely and have a vague image of their surroundings using stereo sound. Our objective is to build a system that can work offline on raspberry pie and use stereo sound feedback systems to guide visually impaired people to walk in an indoor environment. 

Progress Report 4

Updates on Our Progress

  • We designed how the components should be placed on the helmet and assembled them together.
  • We implemented a two-process program for running the main program and the stereo sound program separately.
    • Previously we wanted to use a single python program with two simultaneous processes (Video processing & sound).
    • Through testing we discovered that the two threads could not run together because they require different methods of starting its process.
    • Therefore, we created another program to serve as an audio server, it also has two processes. (Receiving & generating audio) Audio server communicates with the main propgram through a localhost socket connection.
  • The main program divides the image input into four areas, and the stereo sound will respond to the area with the most significant obstacles.

Components and Assembly of the Device

  • Cycling Helmet
  • Depth Camera
  • Raspberry Pi
  • Power Supply
Yichen Zhu with the device

Demo of How the Stereo Sound Works with the Image