9. Image Formats, Files and Cameras
RGB, Gray, Binary, etc.
JPEG, TIFF, PNG, RAW, GIF, BMP, JPEG2000, etc.
CCDs, CMOS, Photographic film, VGA, etc.
10. Cocktail Party Problem
Microphone - 1 Microphone - 2
Separated Source - 1 Separated Source - 2
________________________
Microphone - 1 Microphone - 2
Separated Source - 1 Separated Source - 2
How was this done?
[Source: http://cnl.salk.edu/~tewon/Blind/blind_audio.html via Prof.
Andrew Ng’s ML Class on Coursera ]
11. How was this done?
Cocktail Party Problem Algorithm
[W,s,v] = svd((repmat(sum(x.*x, 1), size(x,1), 1.*x)*x');
[Source: Sam Rowein, Yair Weiss, Eero Simoncelli]
12. Audio – 1D signal
➢
Why Bose systems are so costly?
13. Audio Signals
Why Bose systems are so costly?
Search and read the detailed Quora answer to the above question by
a TA Brad Price of Dr. Bose to the question - “Are Bose products
worth the price?”
– http://qr.ae/Iq1UB
20. Face Morphing Ads
Short face blending video on my website
Feature matching of faces is left to future scope :P
Portrait Professional - http://www.portraitprofessional.com/
Face Morphing Ad
25. Structure from Motion - from 2d
projections
Structure from Motion - 3D from 2D images
26. Application in Medicine
Vision-based Blood Test System gives accurate
results in 15 mins -
http://www.ptgrey.com/news/pressreleases/details.asp?
Vision based Blood Test System
29. Hardware Innovation
Lytro Camera - http://www.lytro.com/camera/
Make the impossible possible. Change your perspective.
Lytro's newest light field capability, Perspective Shift, allows you to
interactively change your point of view in a picture, after you’ve
taken the picture. On a computer or mobile device, you can shift
the living picture in any direction; left, right, up, down and all
around.
Perspective Shift works on light field pictures you've previously taken
and with any new pictures you take. Change your perspective and
see the moment come alive.
30. Hardware Innovation
Like in Matrix and the Tamil film Anniyan,
breakthrough “freeD” sports replay system for
NBC Sunday Night Football to be powered by
Teledyne DALSA cameras
Coming to your TV soon
Femto-Photography
31. AR
Google Glass – Augmented Perspective
Live AR by National Geographic
32. State of the Art Tracking - OpenTLD
Predator Drone - tracking a car from UAV
TLD Tracker Demonstration of Learning
TLD Tracker Human Face
33. 3d model reconstruction from 3D
Camera
Kinect 3D Reconstruction
PrimeSense Capri Demo at Google IO
34. How to start?
Come equipped with good programming skills and read
Wikis in link depths
Several Open Source Projects and Free SDKs to help you
viz. OpenCV, OpenNLP, OpenNI, OpenTLD, Tesseract
for OCR, OpenCL, CUDA for NVIDIA, etc.
Get started with tutorials and examples
Ask exact doubts after having tried solving the problem
(from what I've seen, people do not help you online if
they don't feel you've tried enough) in specific fora and
open fora like StackOverFlow
MatlabCentral for MATLAB specific questions
35. Conclusion
You are now a different person than you were a few hours ago! :)
You'll certainly be more wise and knowledgable after this
workshop, provided you apply what you learn here.
Join the FB Group – I love Computer Vision – if you liked this
presentation.
https://www.facebook.com/groups/visionclass/
Companies working on IP-CV compiled by Prof. David Lowe