Custom Object Tracking With Windows 10 IOT core


Hardware components:
NVIDIA Jetson NVIDIA gpu gtx 750ti
×1
logitech webcam c210
×1
R8326274 01
Raspberry Pi 2 Model B
×1
Software apps and online services:
Vs2015logo
Microsoft Visual Studio 2015
10
Microsoft Windows 10 IoT Core
Microsoft CNTK
Virtual DUB

Custom Object Tracking With Windows 10 IOT core

STORY

Hello

After a long time i decided to post another project as i was really bit by my success in my first project

I wanted to take things further and explore new Human to machine interaction models in IOT space

Hence i came up with this project

Problem that the POC aims to solve:-

Switches are sometimes hard to reach and many a times one is left with the longing for a switch to be much closer to reach .

Voice recognition to switch on/off an appliance is good but not really ok as you are required to continuously address the system in your voice.

Sometimes a person just wishes for a tactical feedback!.

Solution:-

  • Have 2 or more webcams around the room which keep monitoring the room.
  • Create a custom object .(This can be anything custom designed to even one which is designed specifically for room aesthetics )
  • Have the system machine learn what a custom switch looks like and let the location of the switch be decided by the user .
  • Have GPIO pins control relays which switch appliances on /off or interact with services

Due to above solution user is able to dynamically reconfigure physical location and services that the custom switches offer at will at the drop of a hat!

Training method:-

The system borrows from my old project accessible here

and hence forth i have created custom programs and utils for

  • Training with reference to only the object and not the scene
  • Retraining
  • Partial viewing of an object
  • Labeling Utility

Read More: Custom Object Tracking With Windows 10 IOT core




Leave a Comment

© 2015 Powered By Engineering Projects Team, Raspberry Pi Projects

Scroll to top