scene628_segments

BOLD applied on one scene of the CMU-K08 dataset

BOLD features are image descriptors specifically proposed for the detection of texture-less objects. Nevertheless, they also showed to deal with textured objects with a performance comparable to that of state-of-the-art image descriptors. They are efficiently implementable and run in real-time on an off-the-shelf CPU.

 BOLD features were proposed within the following scientific work:

[1] F. Tombari, A. Franchi, L. Di Stefano, “BOLD features to detect texture-less objects”, ICCV 2013 [PDF]

The BOLD algorithm has been developed by CVLab in cooperation with Datalogic Automation within the VIALAB Project, funded by Regione Emilia Romagna. Please note that a patent has been filed covering the invention associated with BOLD features.

A pre-compiled library including the implementation of BOLD as well as of the object detection pipeline described in [1] is available here.

Be bold, go BOLD!