Hello

I'm working on improving the RGBD-SLAM by finding loops in the pose-graph. I'm currently evaluating the vocab_tree with the Calonder-Descriptors to get potentially matching image-pairs which can then be compared.

Since I took some code from the kidnappedrobot stack I'm not sure if I'm using it correctly. So I'd like to know if there is some kind of testing data available. I imagine a set of images (some hundred) together with the pairs of imageids the tree would generate.

Nikolas