Automated Learning Platform for Smart TV Visual Mark Recognition

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing smart TV systems face challenges in automatically identifying and controlling connected devices, especially when service providers' visual marks or UI change, leading to interruptions in the 'one remote' function, requiring manual data collection and human intervention, which is costly and limits scalability.

Innovation Solution

An automated learning platform that stores data related to visual marks and their features, uses a server to map and update these features, and transmits them to secondary devices, enabling automatic recognition and control of connected devices without manual intervention.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a pre-trained machine learning model is used for visual mark identification, then the auto-device setup function can be implemented, but the system cannot adapt when service providers change their visual marks or UI, requiring manual data collection and model retraining

Engineering Contradiction:

Improveadaptability to visual mark changesVSAvoidcomplexity of manual data collection and model retraining

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system performs self-learning by automatically capturing display frames, extracting visual mark regions, and updating the machine learning model without requiring manual data collection or external intervention. The TV autonomously adapts to new visual marks by processing captured images and retraining the model locally

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system continuously captures and stores display frames in advance, maintaining a database of visual mark examples. This preliminary data collection enables the model to be updated seamlessly when visual marks change, without requiring immediate manual intervention or system downtime

Inventive Principle:

Principle #10Preliminary action

2Adaptability or versatility

If the number of supported service providers is increased, then the versatility of the system improves, but the manual effort required for data collection and model training increases significantly

Engineering Contradiction:

Improvenumber of supported service providersVSAvoidtime required for manual data collection and model training

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system automatically learns and adapts to new service providers by capturing their visual marks from displayed content and updating the machine learning model autonomously. This eliminates the need for manual data collection and model retraining for each new provider

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system continuously collects and stores display frames containing visual marks from various service providers in advance. This pre-collection of diverse visual mark data enables the model to be updated efficiently when new providers are introduced, significantly reducing the time required compared to manual training

Inventive Principle:

Principle #10Preliminary action

3Adaptability or versatility

If automated learning is implemented, then the scalability of the system improves, but the initial system complexity and computational requirements increase

Engineering Contradiction:

Improvescalability of the systemVSAvoidcomplexity of automated learning system

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The automated learning system is divided into separate functional modules: frame capture, visual mark region extraction, machine learning model training, and model update. This segmentation allows each component to be optimized independently and simplifies the overall system architecture and maintenance

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs preliminary processing of display frames to extract and store visual mark data in a database before it is needed for model training. This pre-processing and storage of training data reduces the computational burden during model updates and simplifies the real-time operation of the automated learning system

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11343564B2Automated learning platform of a content provider and method thereof

Publication Date: 2022.05.24 SAMSUNG ELECTRONICS CO LTD

AI summary

An automated learning system of a content provider includes a database, an image processing unit, and a server. The database stores data related to visual marks, features of the visual marks, a set of discriminating instances, a position of a region of interest, and pre-defined threshold values. The image processing unit includes a detection module, a determination module, and a feature generation module. The detection module detects frames from a primary display device. The determination module extracts a static visual area, and determines a visual mark. The feature generation module generates discriminating features of the visual mark. The server maps the discriminating features with the stored data, identifies at least one closest visual mark, and transmits the updated visual mark and the discriminating features to secondary display devices.