Antialiasing Neural Networks for AR Object Localization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Deep learning techniques used for image processing in augmented reality experiences, such as virtual try-on (VTO), face challenges with shift variance, leading to perceptible differences in object localization across video frames.

Innovation Solution

The implementation of a system that uses a nail localization engine with deep neural networks configured for antialiasing in both encoder and decoder components, along with a rendering component to provide a VTO experience, addressing shift variance through blur operators that disperse values to neighboring regions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If deep learning networks are used for object localization in augmented reality, then object detection capability is improved, but shift variance causes perceptible differences in localization between video frames

Engineering Contradiction:

Improveobject localization accuracyVSAvoidframe-to-frame consistency

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The patent applies preliminary antialiasing blur operations before object localization in the encoder and before rendering in the decoder. This preprocessing step disperses pixel values to neighboring regions, making the network less sensitive to small shifts between frames and reducing jitter in the output

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent modifies the network architecture by incorporating blur operators that change the parameter distribution of pixel values. By dispersing values to neighboring regions through controlled blurring, the network achieves shift invariance while maintaining localization accuracy across frames

Inventive Principle:

Principle #35Parameter changes

2Reliability

If antialiasing blur operators are applied in encoder and decoder, then shift invariance is achieved, but computational complexity increases

Engineering Contradiction:

Improveshift invarianceVSAvoidnetwork architecture complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies blur operators selectively at specific stages in the encoder and decoder architectures rather than uniformly throughout. This partial application achieves the necessary shift invariance while minimizing unnecessary computational overhead and architectural complexity

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS12327337B2Method and system for object antialiasing in an augmented reality experience

Publication Date: 2025.06.10 LOREAL SA

AI summary

Methods, apparatus and technique embodiments localize objects using antialiasing such as for rendering an object with an effect for a virtual try on (VTO) experience. An example system comprises a nail localization engine including computational circuitry to localize one or more nail objects in an input image of a hand or foot via one or more deep neural networks, wherein the one or more deep neural networks is configured for antialiasing in each of the encoder and decoder components; and a rendering component including computational circuitry to render an output image simulating a nail product or nail service applied to the one or more nail objects, responsive to the localizing by the nail localization engine, to provide a virtual try on (VTO) experience. In an embodiment, respective blur operators in each of the encoders and decoders disperse values to neighboring regions to counteract shift variance.