AR Object Tracking via Server Feature Sets

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing augmented reality (AR) solutions face challenges in maintaining accurate object tracking due to rapid changes in object appearance, lighting conditions, and occlusions, which require frequent updates from object detection servers, leading to network latency and processing power issues.

Innovation Solution

The proposed solution involves generating and transmitting a feature set from the object detection server to the user device, which includes server-generated object feature representations. This feature set enhances object tracking performance by incorporating representations of various object appearances, including unseen instances, thereby improving tracking accuracy under changing conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If frequent updates from object detection server are used to maintain accurate object tracking, then tracking accuracy is improved, but network latency and processing power requirements increase

Engineering Contradiction:

Improvetracking accuracyVSAvoidnetwork latency

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary object detection and extracts feature representations on the server side before transmitting them to the user device. This advance preparation allows the user device to perform lightweight tracking using pre-computed features, reducing the need for frequent server updates and minimizing network latency while maintaining tracking accuracy.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces feature representations as an intermediary between full object detection results and tracking operations. Instead of transmitting complete detection results frequently, the system uses compact feature representations that enable accurate tracking with fewer updates, thereby reducing network latency and processing requirements.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If CNN-based object detection is used to achieve high detection accuracy, then detection precision is improved, but processing power requirements increase

Engineering Contradiction:

Improvedetection precisionVSAvoidprocessing power

Core Design Contradiction:

Measurement precisionVSPower

Solution Approach 1:

The system segments the object detection and tracking tasks between server and user device. The computationally intensive CNN-based detection is performed on the server, while the user device executes lightweight tracking algorithms using pre-computed features from the server, thereby achieving high detection precision without requiring significant processing power on the user device.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The server generates and transmits feature representations that serve as compact copies of the full object detection results. These feature copies enable accurate tracking on the user device without requiring the device to run full CNN detection models, thus reducing processing power requirements while maintaining detection precision.

Inventive Principle:

Principle #26Copying

3Device complexity

If object tracking relies solely on previous frame data, then device complexity is reduced, but adaptability to rapid appearance changes decreases

Engineering Contradiction:

Improvetracking algorithm complexityVSAvoidadaptability to appearance changes

Core Design Contradiction:

Device complexityVSAdaptability or versatility

Solution Approach 1:

The system implements a feedback mechanism where the user device sends tracking results back to the server, and the server responds with updated feature representations when object appearance changes are detected. This feedback loop enables the lightweight tracking algorithm to adapt to rapid appearance changes without increasing device complexity, as the adaptation intelligence resides on the server.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS12236638B2Object tracking and detection in augmented reality applications

Publication Date: 2025.02.25 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)

AI summary

Object tracking for augmented reality involves determining a location of an object in a current frame (10) of a video stream based on object feature representations extracted from different locations in the current frame (10) and object feature representations belonging to a model of the object. The model comprises at least one object feature representation extracted from at least one previous frame (20) of the video stream and at least one object feature representation from a feature set comprising at least one server-generated object feature representation of objects belonging to a same object class as the object.