Product Classification based on SVM and PHOG Descriptor
He Zhang, Zijun Sha · 2013
Summary It is a great need to classify the numerous product into some categories automatically. In this paper, we adopt SVM classifier combined with PHOG (Pyramid of Histograms of Orientation Gradients) descriptors to implement product–image classification. The support vector machine maps the input vectors into a highdimension space, in which a max margin supper hyper plane is set up to classify the samples and PHOG can flexibly represent the spatial layout of local image shape. Experimental results showed the effectiveness of the proposed algorithm.