Indoor scene recognition via multi-task metric multi-kernel learning from RGB-D images

Abstract

The traditional scene analysis mainly focuses on outdoor scene recognition rather than indoor scene understanding. However, with the widespread use of depth cameras, we have a new opportunity to handle the indoor scene recognition problem. In this paper, we propose a multi-task metric multi-kernel learning algorithm that exploits the inter-source similarities and complementarities between color images and depth images to conduct the indoor scene recognition. Specifically, our method utilize multi-task metric learning to learn a Mahalanobis metric for RGB-D images. Multi-task metric learning can extract the common properties from color images and depth images to learn better metrics. Furthermore, the learned metrics are employed to transform features to a correcting feature space for obtaining a better representation. By exploiting multi-kernel learning, our method can leverage multiple feature representations to train a more discriminative classifier. We conduct experiments on NYU Depth Dataset and B3DO Dataset to evaluate the effectiveness of our approach. The experimental results have demonstrated that our proposed method can lead to better indoor scene recognition.