Classification based on SPACT and visual saliency

This paper proposes an efficient approach for object classification. This method bases on bag-of-features classification framework and extends the limits of it. It applies modified spatial PACT as local feature descriptor, which can efficiently catch image patch's characteristic. In order to address the speed bottleneck of codebook creation, Extremely Randomized Clustering Forest is used to create discriminative visual codebook as well as classifier. The prior knowledge stored by the classifier is used to build saliency maps online. The saliency maps can bias the random sampling of sub-windows and improve the speed of classification. Through evaluation on PASCAL 2007 Visual Classification Challenge dataset set, the test results show that this object classification method has many advantages. It has comparable performances to state-of-the-art algorithms with short training and testing times. It has nearly no parameter to tune and it is easy to implement.