[en] Given a video sequence captured from a static viewpoint, the stationary background initialization problem consists in generating a unique image estimating the stationary background of the sequence (i.e. the set of elements which are motionless throughout the sequence). Generating an estimation of the background is helpful, and sometimes crucial for many applications including video surveillance, segmentation, compression, inpainting, privacy protection, and computational photography.
The aim of this talk is to first introduce the background initialization field by presenting the main challenges, some important methods, and the evaluation framework. Second, LaBGen, which emerged as the best method during the Scene Background Modeling and Initialization (SBMI 2015) workshop and IEEE Scene Background Modeling Contest (SBMC 2016), will be presented in depth.