Apple invents advanced Machine Learning-Enabled Object Detection Algorithms to allow users to customize Memoji to a higher degree
This week the US Patent & Trademark Office published a patent application from Apple that relates to the next-generation of Avatar/Memoji customization that will allow a user's to enroll their own custom glasses, hat or other accessory to be integrated into their Memoji through a machine learning process. More specifically, the patent states that "In some embodiments, one or more computer vision (CV) and/or other machine learning (ML)-enabled object detection algorithms may be applied to the enrollment image."
In Apple's patent background they note that Avatars can offer a measure of privacy, while still allowing the users to have control over their online identities. As a means for self-expression, many users may desire to customize their own avatars to show physical characteristics that reflect the actual appearances of the users. The users can also customize their avatars with various accessories, such as clothing items, eyeglasses, headwear, and more, which can reflect the users' individual styles, fashion tastes, and/or real-life appearances.
However, in conventional systems, the various elements of each user's avatar may have to be selected manually from large “libraries” of virtual object elements, which may be tedious and/or present difficulties to the user in terms of locating the virtual object elements that most closely mimic their actual facial expressions, appearances, body language, and/or accessories that they may exhibit in different situations and circumstances. Although users can also select generic (or even randomly-generated) elements to customize their avatars, such generic and/or randomly-generated elements may not accurately reflect the individuality in each user's appearance, style, and/or preferred accessories.
Apple's invention is focused on providing users with a whole new means of creating customized, individualized avatars, wherein each instance of the avatar may be customized to have a particular facial expression, visual characteristics, body language, and/or accessories that may reflect an individual user's appearance in a particular situations and circumstances.
Later in the patent, Apple notes that an individualized avatar can resemble the user that it represents in terms of appearance and fashion. In this way, there is a need for devices, methods, and non-transitory program storage devices (NPSDs) to provide users with more intelligent and automated matching of objects of interest (e.g., eyeglasses frames, etc.), as identified in 2D enrollment images of a user, when determining particular 3D object models to suggest for use during the creation and/or customization of a 3D avatar for the user.
Beyond the basics of future customization being made available to users in Memoji creation that could be appreciated by all, the depth of the patent could only be appreciated by developers and engineers.
For more details, review Apple's patent application # US 20230055013 A1. Apple introduced Memoji in 2018 in iOS 12. Apple's patent published this week was originally filed in late 2021, showing that the invention is to advance Memoji customization in the future.
Apple lists Gabriela Venturini, AV/VR Software Engineer, Technology Development Group as the sole inventor. Ms. Venturini's background includes stints at Walt Disney Imagineering, Activision blizzard and DreamWorks Animation. Ms. Venturini also was an Adjunct Professor at Occidental College where she lectured and guided 15/20 undergraduate students through the laboratory/experimental components of General Chemistry II. Topics included thermodynamics, chemical equilibrium, and reaction kinetics.