On 10/29/2013 07:19 AM, jim vickroy wrote:
On 10/29/2013 5:11 AM, Olivier Grisel wrote:
2013/10/23 j vickroy<jgv.h...@gmail.com>:
On 10/23/2013 10:18 AM, Andreas Mueller wrote:
FYI, the features I would use for the superpixel based approach would be
"color" histogramms (bag of words of channel intensities):
Reshape the images to (-1, 6) so you have lists of pixels (subsample if
they are many), run (MiniBatch)KMeans, and use the cluster-histograms as
features to describe the superpixels.
Then run 1) LinearSVC 2) LinearSVC with AdditiveChi2Sampler 3) SVC 4)
SVC with AdditiveChi2Sampler on the features / labels.
Thanks so much for all of the suggestions on how to approach this!
This is starting to make our Bayesian classifier appear overly (dare I
say it) naive ;-)
For computer vision tasks, the final classifier algorithm is usually
much less important than the way you represent the problem to feed to
it (feature extraction and sub problem decomposition).
Ahhh ... ok. We did not approach the task in that manner. We simply
implemented a Bayesian algorithm, trained it on some Solar images, and
then started using it. As previously noted, it works fairly well but
does have some labeling issues because there is not clear separation
of the multi-channel (wavelength), pixel values so our approach is to
include yet another wavelength (feature) in the analysis. Now that I
have discovered scikit-learn, I am investigating what presumably would
be a more robust (yet simpler) approach.
I wrote at least 3 lines of the tutorial I promised ;)
------------------------------------------------------------------------------
Android is increasing in popularity, but the open development platform that
developers love is also attractive to malware creators. Download this white
paper to learn more about secure code signing practices that can help keep
Android apps secure.
http://pubads.g.doubleclick.net/gampad/clk?id=65839951&iu=/4140/ostg.clktrk
_______________________________________________
Scikit-learn-general mailing list
Scikit-learn-general@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
