AR Depth Estimation Using Pose-Guided Region Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing augmented reality (AR) devices face computational and power constraints in estimating depth due to limited resources, necessitating full-image depth estimation that is computationally demanding.

Innovation Solution

A system that utilizes a 6DOF tracking system to predict the area of interest in the camera's field of view and estimates depth only for this limited region, reducing computational load and power consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If full-image depth estimation is performed, then depth information completeness is improved, but computational load and power consumption increase

Engineering Contradiction:
Improvedepth information completenessVSAvoidpower consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent divides the camera field of view into multiple regions and performs depth estimation only for the region containing the virtual object, rather than processing the entire image. This segmentation approach maintains depth information completeness for relevant areas while significantly reducing computational load and power consumption.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different processing quality levels to different regions of the image. High-quality depth estimation is performed only for the region of interest containing the virtual object, while other regions receive minimal or no depth processing. This local quality approach optimizes the balance between depth information completeness and energy consumption.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If full-image depth estimation is performed, then depth information completeness is improved, but computational resources are overwhelmed

Engineering Contradiction:
Improvedepth information completenessVSAvoidcomputational resources
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the image processing task by identifying and isolating the region containing the virtual object. Depth estimation algorithms are applied only to this segmented region, reducing the computational complexity from processing the entire image to processing a focused subset, thereby preventing computational resource overload.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs depth estimation partially - only for the necessary region containing the virtual object rather than the entire image. This partial action approach provides sufficient depth information for AR rendering while avoiding the excessive computational burden of full-image processing.

Inventive Principle:
Principle #16Partial or excessive action

3Use of energy by moving object

If depth estimation is focused on specific areas, then power consumption is reduced, but depth information coverage decreases

Engineering Contradiction:
Improvepower consumptionVSAvoiddepth information coverage
Core Design Contradiction:
Use of energy by moving objectVSMeasurement precision

Solution Approach 1:

The patent performs preliminary identification of the virtual object's location in the camera field of view before performing depth estimation. This preliminary action allows the system to focus computational resources on the correct region, ensuring that depth information coverage is sufficient for the virtual object while minimizing power consumption.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses the 6DOF tracking data, which provides comprehensive pose information, to guide the depth estimation process. This multi-functional use of tracking data enables the system to maintain accurate depth information coverage for virtual objects while reducing overall computational burden through targeted processing.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP4632692B1Augmented reality guided depth estimation
Publication Date: 2026.04.08 SNAP INC
  • EP4632692B1 patent drawingFigure 1
  • EP4632692B1 patent drawingFigure 2
  • EP4632692B1 patent drawingFigure 3

AI summary

A method for AR-guided depth estimation is described. The method includes identifying a virtual object rendered in a first frame that is generated based on a first pose of an augmented reality (AR) device, determining a second pose of the AR device, the second pose following the first pose, identifying an augmentation area in the second frame based on the virtual object rendered in the first frame, and the second pose, determining depth information for the augmentation area in the second frame, and rendering the virtual object in the second frame based on the depth information.