Image Display Object Recognition via Server-Assisted Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image display apparatuses face challenges in accurately providing information about objects on screen due to timing errors between scheduled and actual program start times, leading to inaccurate information delivery and increased computational load from comprehensive object recognition.
Innovation Solution
An image display apparatus with a network interface and controller that generates screen data, obtains relevant object data from a server, and uses learning models to quickly identify and provide information about objects on screen, reducing computational load and ensuring accurate information delivery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If comprehensive object recognition is performed to identify all objects in the screen, then the accuracy of information provision is improved, but the computational load increases exponentially
Solution Approach 1:
The patent segments the object recognition task by first identifying candidate objects based on pre-stored screen data and time information, then performing recognition only on these candidates rather than all objects in the screen. This divides the comprehensive recognition task into a targeted subset, reducing computational load while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary action by pre-storing screen data and associated object information in advance, then using this pre-processed data to quickly identify relevant objects during actual operation. This preliminary preparation eliminates the need for comprehensive real-time recognition, significantly reducing computational requirements while maintaining high accuracy.
2Measurement precision
If object recognition level is increased to improve recognition accuracy, then the accuracy of information provision is improved, but the computational load increases exponentially
Solution Approach 1:
The patent applies local quality by performing high-accuracy object recognition only on specific candidate objects identified through pre-stored screen data, rather than applying uniform high-level recognition to all objects. This localized approach maintains high recognition accuracy for relevant objects while avoiding the exponential computational cost of comprehensive high-level recognition.
3Quantity of substance
If information on all objects mapped at a certain time is provided, then the completeness of information is improved, but the accuracy of information provision decreases due to timing errors
Solution Approach 1:
The patent uses feedback by comparing pre-stored screen data with actual screen content and using time information to identify and filter relevant objects. This feedback mechanism ensures that only objects actually present in the current screen and relevant to the user's request are included in the information provision, maintaining accuracy while preserving completeness.
Data Source
AI summary
The present disclosure relates to an image display apparatus and an operating method thereof. The image display device according to an embodiment of the present disclosure includes a display; a network interface unit that performs communication through a network; and a controller, wherein the controller generates data for a screen output through the display, when a preset user input is received while a first content is output through the display, obtains data, which corresponds to data of the screen, that is related to an object included in the screen, from a first server through the network interface unit, determines at least one first object related to a position corresponding to the user input, among object included in the screen, based on the data that is related to the object, and outputs a user interface (UI) for the at least one first object through the display. Various other embodiments are possible.