Meet Bob, Idiap’s toolbox latest release

Bob is a signal-processing and machine learning toolbox originally developed by Idiap Research Institute. Upgrading from version 5 to 6, the last release of Bob is offering some important improvements with a focus on video support.

To stay relevant and efficient, a toolbox must be up-to-date.  This is particularly true if the toolbox is the barebones of research work.  Bob is such a Swiss knife for fields related to machine learning, pattern recognition, and signal processing.  To keep the pace with currents needs, Idiap’s researchers just released a new version of Bob, which includes:

-    Improved GIF and PBM reading in
-    Moved to ffmpeg 4 in
-    Implemented Expectation Maximization with multiprocessing in bob.learn.em
-    Added a threading option to the GMM algorithm in
-    Added a cross database testing evaluation with a new command bob pad cross in bob.pad.base
-!37 Keep the order of FrameContainers when saving (breaking change)

And a lot more features and bug fixes. For the full log, please see:

The Bob toolbox is a development joint effort of the Biometrics security & privacy group, the Biosignal processing group, and the Research & development engineers from Idiap.

Project website