You are here

Deep Learning Image Understanding with Subspace Indexing on Grassmann Manifolds

2 July 2015
San Francesco - Via della Quarquonia 1 (Classroom 2 )
In large scale visual pattern recognition applications, when the subject set is large the traditional linear models like PCA/LDA/LPP, become inadequate in capturing the non-linearity and local variations of visual appearance manifold. Kernelized non-linear solutions can alleviate the problem to certain degree, but faces a computational complexity challenge of solving an Eigen problems of size n x n for number of training samples n. In this work, we developed a novel solution by applying a data partition on the BIGDATA training set first and obtain a rich set of local data patch models, then the hierarchical structure of this rich set of models are computed with subspace clustering on Grassmanian manifold, via a VQ like algorithm with data partition locality constraint. At query time, a probe image is projected to the data space partition first to obtain the probe model, and the optimal local model is computed by traversing the model hierarchical tree. Simulation results demonstrated the effectiveness of this solution in capturing larger degree of freedom (DoF) of the problem, with good computational efficiency and recognition accuracy, for applications in large subject set face recognition and image tagging.
Li, Zhu