I've been spending a lot of time looking through OpenCV and other computer vision frameworks. It seems to me that there are a couple of areas where OpenCV could use some improvement:
- Install (seems like many people have a lot of trouble, not that that's surprising)
- Morphological Image similarity (going beyond color histograms and keypoints)
There are others, but I was wondering:
If you could design another version of OpenCV, what features would you look for?
