Augmented Reality Object Recognition Distributed Server Architecture
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Augmented reality systems face inefficiencies in storing and processing large amounts of virtual content, leading to responsiveness and communication issues due to limited mobile device memory and the need for extensive object database filtering, which can be inefficient and costly.
Innovation Solution
A distributed architecture is implemented, grouping descriptors and virtual content into channels and allocating them across multiple servers to balance memory and processing costs, allowing for efficient object matching and load balancing, thereby improving system responsiveness and scalability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a single large object database is used to store all virtual content, then the system can access all objects, but the processing efficiency decreases and memory resources are wasted on objects users are not subscribed to
Solution Approach 1:
The patent divides the single large object database into multiple smaller databases, where each database stores objects associated with specific channels. This segmentation allows the system to access all virtual content while improving processing efficiency by filtering objects at the database level rather than processing all objects uniformly.
Solution Approach 2:
Each database is tailored to store specific subsets of objects relevant to particular channels, creating local optimization for each database. This allows efficient processing for each user's subscribed channels while maintaining the ability to access all content in the system.
2Speed
If all object descriptors are downloaded to mobile devices, then object recognition is faster, but mobile device memory is exceeded and initial download time increases
Solution Approach 1:
The patent extracts only the necessary object descriptors for channels that users are subscribed to, rather than downloading all descriptors. This is achieved by having mobile devices query specific databases based on user subscriptions, extracting only the required data to minimize memory consumption while maintaining fast recognition for relevant objects.
Solution Approach 2:
The system performs partial action by downloading and processing only the subset of objects corresponding to user subscriptions rather than all objects. This partial approach reduces memory usage and download time while maintaining sufficient performance for the intended use case.
3Device complexity
If a centralized database architecture is used, then data management is simpler, but system scalability and load balancing become difficult
Solution Approach 1:
The patent segments the centralized database into multiple distributed databases, each managing specific channels of objects. This segmentation maintains relative simplicity in data management for each node while enabling system scalability through the distributed architecture, where each database can be managed independently and loaded balanced across multiple servers.
Data Source
AI summary
A method for object recognition performed by a computing device of an augmented reality system. The method includes receiving an image from a user, determining channels that a user is subscribed to, and determining a list of servers that host the channels that the user is subscribed to by using groups of channels that are distributed among a cluster of servers. The method further includes selecting, using the computing device, which servers from the list of servers are to be used to query the channels that the user is subscribed to. In addition, the method includes querying, using the computing device, the selected servers for the channels that the user is subscribed to with the image to determine at least one object that matches the image from object databases for the channels that the user is subscribed to. The method also includes retrieving and sending content associated with the at least one object.