Landmark-based inductive model for robust discriminative tracking

The appearance of an object could be continuously changing during tracking, thereby being not independent identically distributed. A good discriminative tracker often needs a large number of training samples to fit the underlying data distribution, which is impractical for visual tracking. In this paper, we present a new discriminative tracker via the landmark-based inductive model (Lim) that is non-parametric and makes no specific assumption about the sample distribution. With an undirected graph representation of samples, the Lim locally approximates the soft label of each sample by a linear combination of labels on its nearby landmarks. It is able to effectively propagate a limited amount of initial labels to a large amount of unlabeled samples. To this end, we introduce a local landmarks approximation method to compute the cross-similarity matrix between the whole data and landmarks. And a soft label prediction function incorporating the graph Laplacian regularizer is used to diffuse the known labels to all the unlabeled vertices in the graph, which explicitly considers the local geometrical structure of all samples. Tracking is then carried out within a Bayesian inference framework where the soft label prediction value is used to construct the observation model. Both qualitative and quantitative evaluations on 65 challenging image sequences including the benchmark dataset and other public sequences demonstrate that the proposed algorithm outperforms the state-of-the-art methods.