Preserving the semantic meanings by keeping the contents of the original image as it is without disturbing its nature is the process of ‘Image Retargeting’. It’s wide aim is to display the image on any arbitrary size of display device irrespective of its aspect ratio or resolution. When the image is to be fitted in the target resolution, certain unimportant regions need to be deleted. The existence method predicts the pixel-wise importance in a bottom-up manner. In contrast, the proposed algorithm uses the top-down approach. This estimation is fused by ‘Classification guided Fusion Network (CFN). This feature is widely applied in the 3-D image retargeting which fuses the left and right eye images which are differently designed. Semantic collage the images by fusion.