Weighted Deformable Part Model for Robust Human Detection

Due to human pose articulation, variation in human shapes and appearances, especially occlusion between human and objects, one challenging problem in human detection is detect partially or completely occluded humans. In this paper, we propose a novel human detection approach capable of handling partial occlusion by combining the deformable part model and mean-shift algorithm. Since the training of root and part filters in deformable part model is separated, we construct the occlusion confidence map through root filter. Then mean-shift algorithm is used to segment the confidence map. The segmented portion of the window with a majority of negative response is referred to the partial occlusion. By constructing the relationship between deformable part model and the segmented sliding window, the negative influence of occlusion part is significantly weakened. This method dramatically decreases the effect of HOG feature of any occluded part during the decision-making process. Experimental results on the well-known INRIA data set demonstrate the effectiveness of the proposed method to solve the occlusion problem.