Dynamic scene classification: Learning motion descriptors with slow features analysis

In this paper, we address the challenging problem of categorizing video sequences composed of dynamic natural scenes. Contrarily to previous methods that rely on hand-crafted descriptors, we propose here to represent videos us-ing unsupervised learning of motion features. Our method encompasses three main contributions: 1) Based on the Slow Feature Analysis principle, we introduce a learned lo-cal motion descriptor which represents the principal and more stable motion components of training videos. 2) We integrate our local motion feature into a global cod-ing/pooling architecture in order to provide an effective sig-nature for each video sequence. 3) We report state of the art classification performances on two challenging natural scenes data sets. In particular, an outstanding improvement of 11 % in classification score is reached on a data set intro-duced in 2012. 1