A simple wrapper for FLANN, Fast Library for Approximate Nearest Neighbors. It has an interface similar to the NearestNeighbors package API.
Clone package from this repository and build it.
julia> Pkg.clone("https://github.com/wildart/FLANN.jl.git")
julia> Pkg.build("FLANN")
Depending on the version of your operation system, you'll be prompted to install a FLANN binary or it is going be built from the sources.
using Distances
using FLANN
X = readdlm(Pkg.dir("FLANN", "test", "iris.csv"), ',')
v = X[:, 84]
k = 3
r = 10.0
idxs, dsts = nearest(X, v, k, FLANNParameters())
# or
t = flann(X, FLANNParameters(), Minkowski(3))
inds, dists = nearest(t, v, k)
# or
idxs, dsts = inball(t, v, r)
# Do not forget to close index!
close(t)
- Documentation