Section: New Software and Platforms

Yael library

Participants : Matthijs Douze [correspondant] , Herve Jegou [TEXMEX Team Inria Rennes] .

Yael [14] is a library with Matlab and Python bindings providing optimized (multi-threaded, Blas/Lapack, low level optimization) implementations of functions useful in vision and machine learning such as k-means, GMM, exact nearest neighbor search and Fisher vector computation.

In 2014, it was extended to include a generic inverted file implementation, that can accomodate any type of signature that refines the similarity computation between documents. The Fisher vector computation code was also optimized.