KEMBAR78
Hand Gesture Recognition Using OpenCV Python | PPTX
HUMAN COMPUTER INTERACTION USING HAND
GESTURES BY AFFORDABLE ALTERNATIVE TO
DEPTH CAMERA
INTRODUCTION
For the last three decades we are stuck at the
tradtional mouse keyboard setup . Recently with
the introduction of touch screen in smart phones
and the emergance of Augmented Reality and
Virtual Reality devices and state of the art sensors
like Leap Motion and Kinect we are taking a leap
into the future of human computer interaction
THE TIMELINE
1977
THE FIRST MASS
MARKETED PERSONAL
COMPUTER APPLE II
1982
THE FIRST PC
TO USE MODERN
TRACKBALL BASED
MOUS
2015
MODERN PC STILL
USE THE THREE
DECADE OLD MOUSE
KEYBOARD SETUP
WHY CHANGE?
With The introduction of Virtual Reality and
Augmented Reality deivecs this tradtional mouse
keyboard setup is of no use . We just can't intteract
with a Virtual Reality System with a mouse and
keyboard
THE OBJECTIVE
Our Objective is to built a low cost system with
help of low cost hardware and open source
software to provide robust and accurate hand
gesture recognition and tracking.
THE INSPIRATION
THE CHALLENGES - HARDWARE
KINECT SENSOR
Rs 14,900
Depth Camera
LEAP MOTION
Rs 8,900
Motion Sensor
LEAP MOTION
Rs 12,000
3D Sensor
OUR APPROACH - HARDWARE
IR ILLUMINATOR
Rs 120
For IR Illumination
WebCam
Rs 1000
For Vision Based Motion Sensing
And Gesture recognistion
OUR APPROACH SOFTWARE
OPEN COMPUTER VISION LIBRARY WITH PYTHON
THE PROJECT
 To build a hand gesture recognition system that
doesn’t get affected by external factors such as
light , distance and movements.
 To build a system that can recognize with high
accuracy.
 To build a system that can help us interact with
a computer with ease.
HOW WE DO IT
HARDWARE SOFTWARE
WHY USE THIS HARDWARE
SETUP
BUILD THE CORE SOFTWARE
TRAINING THE SVM MODEL AND
CHECKING THE ACCURACY
THE WORK FLOW
DETECTIN
G AND
PREDICTI
NG THE
HAND
POSE
EXPERIMENTS AND RESULTS
HAND CONTROLLER
VIRTUAL WHEEL
MULTI TOUCH
LIMITATIONS AND IMPROVEMENT
 LIMITATION
 FPS limitations cannot detect fast moving hand
 The System makes an assumption that the hand is
the closest and the brightest object to the camera
LIMITATION AND IMPROVEMENT
 IMPROVEMENTS
 Use a common Light source
 Use a IR illuminator and a band pass filter to filter
out unnecessary background objects
 Use of IR illuminator to estimate depth
 Use of High FPS webcam as image source
 Using a depth camera to understand the scene
 Using machine learning techniques to estimate
hand pose
FUTURE IMPROVEMENTS
• FUTURE SCOPE
• Using more sophisticated machine learning
techniques
• Try to build a complete product .
• Miniaturization of the whole system.
• Running the device with least power.
THANK YOU

Hand Gesture Recognition Using OpenCV Python

  • 1.
    HUMAN COMPUTER INTERACTIONUSING HAND GESTURES BY AFFORDABLE ALTERNATIVE TO DEPTH CAMERA
  • 2.
    INTRODUCTION For the lastthree decades we are stuck at the tradtional mouse keyboard setup . Recently with the introduction of touch screen in smart phones and the emergance of Augmented Reality and Virtual Reality devices and state of the art sensors like Leap Motion and Kinect we are taking a leap into the future of human computer interaction
  • 3.
    THE TIMELINE 1977 THE FIRSTMASS MARKETED PERSONAL COMPUTER APPLE II 1982 THE FIRST PC TO USE MODERN TRACKBALL BASED MOUS 2015 MODERN PC STILL USE THE THREE DECADE OLD MOUSE KEYBOARD SETUP
  • 4.
    WHY CHANGE? With Theintroduction of Virtual Reality and Augmented Reality deivecs this tradtional mouse keyboard setup is of no use . We just can't intteract with a Virtual Reality System with a mouse and keyboard
  • 5.
    THE OBJECTIVE Our Objectiveis to built a low cost system with help of low cost hardware and open source software to provide robust and accurate hand gesture recognition and tracking.
  • 6.
  • 7.
    THE CHALLENGES -HARDWARE KINECT SENSOR Rs 14,900 Depth Camera LEAP MOTION Rs 8,900 Motion Sensor LEAP MOTION Rs 12,000 3D Sensor
  • 8.
    OUR APPROACH -HARDWARE IR ILLUMINATOR Rs 120 For IR Illumination WebCam Rs 1000 For Vision Based Motion Sensing And Gesture recognistion
  • 9.
    OUR APPROACH SOFTWARE OPENCOMPUTER VISION LIBRARY WITH PYTHON
  • 10.
    THE PROJECT  Tobuild a hand gesture recognition system that doesn’t get affected by external factors such as light , distance and movements.  To build a system that can recognize with high accuracy.  To build a system that can help us interact with a computer with ease.
  • 11.
    HOW WE DOIT HARDWARE SOFTWARE
  • 12.
    WHY USE THISHARDWARE SETUP
  • 13.
  • 14.
    TRAINING THE SVMMODEL AND CHECKING THE ACCURACY
  • 15.
  • 16.
  • 17.
  • 18.
  • 19.
  • 20.
  • 21.
    LIMITATIONS AND IMPROVEMENT LIMITATION  FPS limitations cannot detect fast moving hand  The System makes an assumption that the hand is the closest and the brightest object to the camera
  • 22.
    LIMITATION AND IMPROVEMENT IMPROVEMENTS  Use a common Light source  Use a IR illuminator and a band pass filter to filter out unnecessary background objects  Use of IR illuminator to estimate depth  Use of High FPS webcam as image source  Using a depth camera to understand the scene  Using machine learning techniques to estimate hand pose
  • 23.
    FUTURE IMPROVEMENTS • FUTURESCOPE • Using more sophisticated machine learning techniques • Try to build a complete product . • Miniaturization of the whole system. • Running the device with least power.
  • 24.