implement texton regions and clustering

RockStarCoders / alienMarkovNetworks

Using MRFs and CRFs for computer vision problems.

21 stars 9 forks source link

implement texton regions and clustering #35

Open jsherrah opened 10 years ago

jsherrah commented 10 years ago

We have been working directly with continuous-valued feature vectors. The famous work of Shotton 2006 on textonBoost builds higher-level features on top of whatever features you are using. The steps are:

cluster your features using k-means
create feature a feature for a random cluster and random relative rectangle.

This gets used in adaBoost, but for now the task is just to implement the clustering and rect. feature calcuation given a set of continuous features.

jsherrah commented 10 years ago

someone here has started working on this, so maybe hold off for the mo?

amb-enthusiast commented 10 years ago

Sure - good to see you promoting gitHub working.

I'll be interested to see the results for k-means clustering on features.

jsherrah commented 10 years ago

someone here has stopped working on this :)