A digital magazine server presents content, which may include images, retrieved from various sources to a user. To improve presentation of images while allowing modification of images, the digital magazine server identifies feature points in an image and identifies regions of the image including the feature points. Groups of regions are generated based at least on the number of feature points in each region and the location of each region. Based on information associated with the groups, such as the location of various groups and the aspect ratio of various groups, one or more groups are selected and associated with the image. The selected groups may identify regions of the image including an object or not including an object, providing information for modifying the image without obscuring objects in the image.