Background modeling by shifted tilings of stacked denoising autoencoders Jorge Garc´ ıa-Gonz´ alez, Juan M. Ortiz-de-Lazcano-Lobato, Rafael M. Luque-Baena, and Ezequiel L´ opez-Rubio Department of Computer Languages and Computer Sciences. University of M´ alaga. Bulevar Louis Pasteur, 35. 29071 M´ alaga. Spain. {jorgegarcia,jmortiz,rmluque,ezeqlr}@lcc.uma.es Abstract. The effective processing of visual data without interruption is currently of supreme importance. For that purpose, the analysis sys- tem must adapt to events that may affect the data quality and maintain its performance level over time. A methodology for background model- ing and foreground detection, whose main characteristic is its robustness against stationary noise, is presented in the paper. The system is based on a stacked denoising autoencoder which extracts a set of significant features for each patch of several shifted tilings of the video frame. A probabilistic model for each patch is learned. The distinct patches which include a particular pixel are considered for that pixel classification. The experiments show that classical methods existing in the literature ex- perience drastic performance drops when noise is present in the video sequences, whereas the proposed one seems to be slightly affected. This fact corroborates the idea of robustness of our proposal, in addition to its usefulness for the processing and analysis of continuous data during uninterrupted periods of time. Keywords: Background modeling · deep learning · autoencoders 1 Introduction Visual pieces of information such as images or video sequences are massively generated and used nowadays. Therefore, reliable and efficient ways to process that kind of data are needed more than ever. Video surveillance remains a very active field in the area of artificial vision, due to the fact that some demanding tasks have not been addressed adequately yet, as it is the case of background modeling, which consists of deciding whether an object of an image belongs to the scene foreground or background. Robustness is a key feature which foreground detection algorithms must present. They should work continuously and they have to be prepared to cope with events which make the background characteristics vary. A change in the weather conditions in outdoor environments or lightning variations in indoor en- vironments may compromise the reliability of moving object detection. There- fore, the algorithm performance must be kept at an acceptable level not only