Hybrid Machine Vision Client Terminal for Distributed Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine vision technologies face challenges in achieving high recognition rates and processing speeds, especially in mobile terminals due to hardware limitations and varying user environments, and are also hindered by on/off-line conditions and network response times when performed on servers.
Innovation Solution
A client terminal with a communication unit, recognition unit, determination unit, and control unit that distributes object recognition tasks between the client terminal and server, using artificial intelligence to set parameters and integrate results for improved performance and accuracy, regardless of hardware performance or user environment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If machine vision is performed in a mobile terminal, then real-time object recognition is achieved, but processing speed and recognition rate are degraded due to hardware performance limits
Solution Approach 1:
The patent divides the machine vision processing into two segments: the mobile terminal performs initial object recognition using its hardware, and the server performs verification and refinement. This segmentation allows the mobile terminal to achieve real-time processing while the server compensates for accuracy limitations, resolving the contradiction between speed and recognition rate.
2Reliability
If machine vision is performed on a server, then recognition accuracy is improved, but response speed degrades due to network transmission and server response time
Solution Approach 1:
The mobile terminal performs preliminary object recognition before transmitting images to the server. This preliminary action filters out cases that can be handled locally, so only complex or uncertain cases are sent to the server, thereby maintaining fast response speed while improving accuracy for difficult cases.
3Reliability
If machine vision is performed on a server, then object recognition accuracy is improved, but service availability is reduced due to on/off-line environmental states
Solution Approach 1:
The mobile terminal is equipped with machine vision capabilities to perform self-service object recognition when the server is unavailable. The terminal can independently recognize objects using its own hardware and algorithms, ensuring service continuity and adaptability to offline environments while maintaining the option to use the server for enhanced accuracy when available.
Data Source
AI summary
A client terminal according to an embodiment of the present invention includes: a communication unit receiving a request for object recognition; a recognition unit performing the object recognition through machine vision; a determination unit determining devices that will perform distributed object recognition; and a control unit setting parameters affecting the object recognition, and performing learning for the object recognition through artificial intelligence on the basis of the set parameters, wherein the control unit controls the determination unit to determine the devices by taking into account the parameters when the request for object recognition is received, and controls the recognition unit to integrate results obtained from the devices performing the distributed object recognition so as to perform the object recognition.