Vertices for contours duplicated with small targets

[king-shak]

If a target is too small or the corners are too round the pipeline will detect multiple points in one corner. So it normally just makes duplicates of the vertices - it doesn't come up with random points. Because of this, we can solve this issue by adding k-means clustering to the pipeline for the vertices. You essentially pass in the points, how many vertices (clusters) you want and a few other parameters. So it will basically return the average of any duplicate points in one area.
This also means we need to add a slider in the web interface to specify how many vertices are desired.
This solution has already been tested, just hasn't been implemented in the pipeline yet. Should come in the next commit.
Some docs on the method: https://docs.opencv.org/2.4/modules/core/doc/clustering.html#kmeans