Dynamic Texture Recognition using a Hybrid Generative-Discriminative Approach with Hidden Markov Models and Support Vector Machines
Samr Ali, Nizar Bouguila · 2019
Dynamic textures (DT) constitute of objects characterized by stationary properties in time such as how leaves move in a windy day. Classification of DTs has made tremendous impact in various domains such as video synthesis and segmentation. In this paper, we propose the use of Fisher kernels with Dirichlet based and Beta-Liouville (BL) based hidden Markov models (HMM) for DT recognition. Experiments demonstrate promising results on the DynTeX Alpha dataset using the proposed generative-discriminative approach. To the best of our knowledge, this is the first application of Dirichlet and BL HMMs to DT recognition.