Abstract
In the domain of video-based image set classification, a considerable advance has been made by modeling each video sequence as a linear subspace, which typically resides on a Grassmann manifold. Due to the large intra-class variations, how to establish appropriate set models to encode these variations of set data and how to effectively measure the dissimilarity between any two image sets are two open challenges. To seek a possible way to tackle these issues, this paper presents a graph embedding multi-kernel metric learning (GEMKML) algorithm for image set classification. The proposed GEMKML implements set modeling, feature extraction, and classification in two steps. Firstly, the proposed framework constructs a novel cascaded feature learning architecture on Grassmann manifold for the sake of producing more effective Grassmann manifold-valued feature representations. To make a better use of these learned features, a graph embedding multi-kernel metric learning scheme is then devised to map them into a lower-dimensional Euclidean space, where the inter-class distances are maximized and the intra-class distances are minimized. We evaluate the proposed GEMKML on four different video-based image set classification tasks using widely adopted datasets. The extensive classification results confirm its superiority over the state-of-the-art methods.