I am trying to build a Distributed KD tree using map-reduce. Description of Distributed KD tree can be found out here Dkd-Tree
I have a feature vector of images having dimension 20. I have to build distributed kd tree according to above link, also Check out this image Kdtree
I have set of millions of images. So what method I can use to build the top of the tree (2nd part of the image) ? I am confused in the image distribution among the various nodes.
If the tree is built HDFS in first map-reduce operation,then how can I access it in next map-reduce operation?
来源：https://stackoverflow.com/questions/11009714/building-distributed-kd-tree-using-map-reduce