Is the "normalized correlation" a standard well-known procedure for clustering? Could you give an example to explain this?

Can an alphabet be extended by adding some "letters" based on the original letters but modified in some way ? (affine transform, color change to simulate a different lighting ...). Or is it not necessary : those info are already here, if the original alphabet is random enough

They state that multiple detections around a point can be used as a measure of confidence (which has worked well for my project), but how do they use this in their activation maps?

Edited by author 11-30-2004 12:39 PM
An earlier version of this paper was presented at ECCV 02: http://l2r.cs.uiuc.edu/~danr/Papers/vrelations.pdf

There's an interesting observation in the earlier paper that's been omitted from this version (at least, I didn't see it). They found that omitting singleton vocabulary items - anything oddball enough that it didn't cluster - degraded the performance of their detector.

This suggests something fundamental about the appropriate model for an object category (such as car, motorbike, face). Looking for averaged prototypes or even for a few elements that are common throughout all examples may not be the best approach. It may be that instead there are some shared elements in one subset and different shared elements in other subsets. For example, if your training set had just one example of a car from the 50s with fins, that car might have few detectable features in common with a contemporary Hyundai or Nissan. But the fins (and maybe some other characteristics) might create features that allow you to detect more cars from that era than if you were to drop out those features just because they're uncommon in your training set.

I'd be interested to see if the automatic selected features would be useful/good for the parts model of a pictorial structure. If so, it would help to alleviate a downside of that method - picking appropriate parts/connection models.