Hello
I'm currently working on detecting similar images using descriptors like Sift, Surf,... I extract the descriptor vectors (currently Sift with 128 dimensions) and want to create a flann-Index for all descriptors I found in my set of images. The tree is created via
tree = new cv::flann::Index(all_descs,cv::flann::KDTreeIndexParams(8));
where all_descs contains the descriptors in all images and is of size (67.000,128). This works well if I create the tree for about 60.000 descriptors. In fact, the largest tree I could train had a size of 65339. And I just realized that this number is close to 65535. I guess that it is possible to use more than 2^16 elements, could anyone tell me how I can create larger trees?