Flann vs. Brute Force matcher

Andrei CACIO

Lately I've been reading about CUDA, feature detection, Flann based matcher, Brute force matcher etc. I've recently developed a demo with GPU Brute Force matcher and the computation is pretty fast. I've also read that the Flann based matcher is much much faster but give low performance in some scenarios. ( I didn't test my demo with flann )

So my question is: For a very very very large database of descriptors which matcher should I use to get good results?

If you suggest brute force matcher, is there a way I could make it even more faster? (multi threading etc.)

If you suggest flann, is there a way to make it more accurate? (and keep the speed)?

MysteryGuy

BruteForceMatcher is a greedy matcher, which compares each pair of descriptors, so FlannMatcher should be faster on large databases. On the other hand, if you use special metrics, you need to use BruteForceMatcher

Collected from the Internet

Please contact [email protected] to delete if infringement.

edited at
0

Comments

0 comments
Login to comment

Related