Image Segmentation And Contextual Modeling For Object Recognition