Automated image augmentation using a virtual character

A method for use by an image processing system including a computing platform having a hardware processor and a system memory storing an image augmentation software code, a three-dimensional (3D) poses library including a plurality of poses, and a 3D body shapes library including a plurality of body shapes, the method comprising:

• obtaining, by the image augmentation software code executed by the hardware processor, an image including a real-world environment image portion corresponding to a real-world environment, a marker image portion corresponding to a marker in the real-world environment, and a human image portion corresponding to a human in the real-world environment;
• providing, by the image augmentation software code executed by the hardware processor, at least one portion of the image including the human image portion as an input to the 2D pose estimation software;
• receiving from the 2D pose estimation software, by the image augmentation software code executed by the hardware processor, a 2D pose data generated based on the at least one portion of the image including the human image portion;
• identifying, by the image augmentation software code executed by the hardware processor, a 3D pose and a 3D body shape corresponding to the at least one portion of the image including the human image portion, using the 2D pose data, the plurality of poses in the 3D poses library and the plurality of body shapes in the 3D body shapes library, wherein the 3D pose is initialized to fit inside a bounding box of the marker;
• and
• generating, by the image augmentation software code executed by the hardware processor, a virtual character having the identified 3D pose and the identified 3D body shape.