Abstract

In multi-view facial expression recognition, discriminative shared Gaussian process latent variable model (DS-GPLVM) gives better performance than that of linear and nonlinear multi-view learning-based methods. However, Laplacian-based prior used in DS-GPLVM only captures topological structure of data space without considering the inter-class separability of the data, and hence the obtained latent space is suboptimal. So, we propose a multi-level uncorrelated DS-GPLVM (ML-UDSGPLVM) model which searches a common uncorrelated discriminative latent space learned from multiple observable spaces. A novel prior is proposed, which not only depends on the topological structure of the intra-class data, but also on the local-between-class-scatter-matrix of the data onto the latent manifold. The proposed approach employs an hierarchical framework, in which, expressions are first divided into three sub-categories. Subsequently, each of the sub-categories are further classified to identify the constituent basic expressions. Experimental results show that the proposed method outperforms state-of-the-art methods in many instances.

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.