Image processing system to align a target object in a target object image with an object model

An image processing system for use on images containing visual information depicting a target object, comprising:

• one or more processors;
• one or more computer readable memories;
• a plurality of base object models stored on the one or more computer readable memories, each base object model of the plurality of base object models representing a respective type of target object of a plurality of different types of target objects on which image processing can take place, each different type of target object corresponding to a different make, model, and year of target objects, and each base object model including a respective plurality of landmarks defining respective particular physical locations associated with the respective type of target object;
• a plurality of sets of correlation filters, stored on the one or more computer readable memories, wherein each set of correlation filters respectively corresponds to a different base object model of the plurality of base object models, each correlation filter included in the each set of correlation filters is trained on images corresponding to the different base object model, and at least one different correlation filter of the each set of correlation filters is associated with each landmark of the respective plurality of landmarks included in the different base object model;
• and
• a routine stored on the one or more computer readable memories for execution on the one or more processors, that when executed on the one or more processors:
• processes one or more images that depict a target object to determine a particular type of the target object;
• selects, based on the determined, particular type of the target object, a particular base object model from the plurality of base object models stored within the one or more computer readable memories associated with the target object;
• selects, based on the particular base object model, one or more correlation filters included in a particular set of correlation filters of the plurality of sets of correlation filters, the particular set of correlation filters respectively corresponding to the particular base object model;
• scans a target object image of the one or more images depicting the target object using the selected one or more correlation filters to detect, within the target object image, one or more landmarks of the plurality of landmarks included in the particular base object model;
• and
• uses the one or more detected landmarks within the target object image to align the particular base object model and the target object depicted within the target object image.