Nonparametric Feature Matching Based Conditional Random Fields for Gesture Recognition from Multi-Modal Video

摘要: We present a new gesture recognition method that is based on the conditional random field (CRF) model using multiple feature matching. Our approach solves labeling problem, determining categories and their temporal ranges at same time. A generative probabilistic formalized probability densities are nonparametrically estimated by matching input features with training dataset. In addition to conventional skeletal joint-based features, appearance information near active hand in an RGB image exploited capture detailed motion of fingers. The likelihood function then used as unary term for our CRF model. smoothness also incorporated enforce coherence solution. Frame-wise results can be obtained applying efficient dynamic programming technique. To estimate parameters proposed model, we incorporate structured support vector machine (SSVM) framework perform learning large-scale datasets. Experimental demonstrate provides effective challenging real By scoring 0.8563 mean Jaccard index, has state-of-the-art track 2014 ChaLearn Looking People (LAP) Challenge.