I would like to fit to a specific shape in an image, let's say something similar to an elliptical shape with some variation. Right now I've annotated some images by placing some landmark points along this shape, and the point distribution model will model the variation in the location of these points. Then I train an AAM on these labeled landmarks. But in reality I don't really care about the exact location of specific points, just the shape that they define. It seems like I now have unnecessary variation coming from different sets of landmarks that define the same shape, but with the landmarks at a bit different locations on the shape.
Is there something in Menpo that I should be using instead, or would this still be a good approach and I just need to be very careful with labeling the landmarks on my shape consistently?
I would like to fit to a specific shape in an image, let's say something similar to an elliptical shape with some variation. Right now I've annotated some images by placing some landmark points along this shape, and the point distribution model will model the variation in the location of these points. Then I train an AAM on these labeled landmarks. But in reality I don't really care about the exact location of specific points, just the shape that they define. It seems like I now have unnecessary variation coming from different sets of landmarks that define the same shape, but with the landmarks at a bit different locations on the shape.
Is there something in Menpo that I should be using instead, or would this still be a good approach and I just need to be very careful with labeling the landmarks on my shape consistently?