[email protected]
KTH - Royal Institute of Technology Sweden
Mårten Björkman
A CUDA implementation of SIFT for NVidia GPUs (1.2 ms on a GTX 1060)
Celebrandil
