Domain adaptation losses

A non-transitory computer readable medium that embodies instructions that cause one or more processors to perform a method comprising:

• (a) selecting a 3D model corresponding to an object;
• (b) generating a first 2D image of the 3D model superimposed on a first background image in a first pose, with a centroid of the 3D model being located at a first position relative to the first background image;
• (c) generating a second 2D image of the 3D model superimposed on the first background image, with the centroid of the 3D model being located at the first position relative to the first background image, the 3D model having a second pose in the second 2D image different from the first pose;
• (d) determining, using an algorithm, a first location of a first feature on the 3D model in the first 2D image and a second location of a second feature on the 3D model in the second 2D image;
• (e) calculating a difference based on the first location and the second location;
• (f) adjusting parameters representing the algorithm based on the calculated difference;
• (g) iterating steps (d) to (f) at least twice;
• and
• (h) storing, in a memory, parameters representing the algorithm, the parameters causing the difference in (e) to be the highest among the iteration or higher than or equal to a threshold.