About
Top Posts
- Some simple ways to speed up convnet a little
- Some thoughts about deep learning criticism
- Geoffrey Hinton on max pooling (reddit AMA)
- Finds in arxiv, January.
- "get nan or inf error" in cuda-convnet - possible fix
- December finds in #arxiv
- November finds in #arxiv and NIPS 2013
- Deriving Gibbs distribution from stochastic gradients
- Meaning of conditional probability
- Finds in arxiv, october
Blogroll
downloads
My places
Old homepage
People
Symbian
Video
- 2d 3d 3d imaging Anathem Android arxiv Augmented Reality Augmented Reality advertisng bionic eye brain-computer interface bundle adjustment C++ camera carbon nanotubes cnn Coding Coding AR compressed sensing computer vision convnet convolutional network CPU cuda cuda-convnet Cyborg Deep Learning FAST feature detection FFT fiduciary marker fun Games Google History hologram image processing Image Registration iPhone L1 minimization machine learning Maemo Markerless Markerless tracking Math Mobile MSER N95 Neal Stephenson neural network Nintendo DSi Nokia Nokia 5800 Nokia N95 nonlinear optimization optimization pose estimation Rainbow Ends RjDj robust statistics Samsung i8510 Samsung INNOV8 scale space Science SIFT SURF Symbian Symbian Signed technological singularity Technology Tower Defense Usability Vernor Vinge videoglasses Vuzix Wrap 920AV xkcd
Twitter
- RT @ctrlcreep: Airplane mode is useful when traversing the earth's secret corridors, as otherwise mole men and machine elves will hunt you… 2 months ago
- @djstrouse SCP-033 scp-wiki.wikidot.com/scp-033 3 months ago
- Known Unknown: Statistical Learning Unknown Known: Game Theory Unknown Unknown: Black Swan around the corner 3 months ago
- RT @Mellow_Pascoe: Have a lovely day 😊 https://t.co/TZTDA4XLGU 4 months ago
- RT @docmilanfar: Mantra https://t.co/ewX5Hm0N2P 4 months ago
Archives
- December 2014 (2)
- November 2014 (1)
- February 2014 (1)
- January 2014 (2)
- December 2013 (1)
- November 2013 (2)
- October 2013 (2)
- September 2013 (1)
- May 2013 (1)
- January 2013 (1)
- December 2012 (1)
- March 2012 (1)
- January 2012 (1)
- November 2011 (1)
- September 2011 (1)
- August 2011 (1)
- July 2011 (1)
- May 2011 (2)
- April 2011 (5)
- November 2010 (3)
- September 2010 (1)
- April 2010 (1)
- February 2010 (4)
- January 2010 (6)
- December 2009 (4)
- November 2009 (4)
- October 2009 (7)
- September 2009 (2)
- August 2009 (4)
- July 2009 (8)
- June 2009 (5)
- May 2009 (1)
- April 2009 (3)
- March 2009 (11)
- February 2009 (10)
- January 2009 (12)
- December 2008 (10)
- November 2008 (12)
- October 2008 (10)
Meta
Mirror Image 
Dear Sergey Ten,
I would like to start by thanking you for sharing the work you have put in the SMMT library under an open source license.
I downloaded the library some weeks ago and started experimenting with the example. As part of my university project, I am building an AR application that runs on Nokia N95 that needs to detect 3 different markers and display the related model with the marker. In relation to this, I would like to ask you the following questions:
1. What program can i use and in what format should i export my 3D models? In other words, how do I export a 3D model in a format that OpenGL ES can then load it from?
2. How can I use three of the 6 markers you provided to render each 3D model? I need to use 3 markers in total, where each marker should be displaying a different 3D model.
I would also like to show my appreciation for the interesting articles you put on the mirror2images blog.
Looking forward to hear from you.
Regards,
Aaron Buhagiar
Hello,
I am a student doing my MS. Stumbled upon your work while looking for works on MSER based image registration. My question is after having extracted the MSER how do you track the features? Having obtained the extremal region and the seed of the level set. What features of the region are used to track? And how do you solve for the affine transformation the base image has gone through?
It will be great help if you could point me in the right direction.
Thanking you sincerely
Vijay