starkdg / mvptree Goto Github PK

Current MVPTree implementaion seems implemented in a way that it
requires a bunch of modifications to be used together with pHash, in
particular it uses double as value type and plain difference as distance, which
seems totally incompatible with pHash hashes and their hamming distance

Using a bit of C++ templates I think I could generalize MVPTree in not much
time but I might not have time to maintain a fork by myself in the future, if
I do prepare a a pull request for generalized MVPTree would you review it and
merge it, or the incompatible typing has been chosen on purpose?