Object Recognition for Video Recommendation via Database Intermediary
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current moving image distribution services face challenges in automatically associating specific objects in video content with relevant recommendations, as existing object recognition techniques can only identify abstract information like 'vehicle' or 'person' but not detailed information like 'type of vehicle' or 'name of person', making it difficult to provide appropriate recommendations to users.
Innovation Solution
An information providing device that includes an object recognizing unit to extract specific objects from moving image content, a retrieving unit to retrieve similar images, and a transmitting unit to send relevant recommendations to the user's terminal device, using a similar image retrieval device to find matching images and obtain related information for display.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If object recognizing technique is used to extract keyword corresponding to reproducing time, then automation is improved, but measurement precision deteriorates
Solution Approach 1:
The patent introduces a database storing detailed information about objects (including images, names, and descriptions) as an intermediary between the object recognition technique and the keyword extraction. The recognition technique extracts abstract object information, which then queries the database to retrieve detailed specific information, thereby resolving the precision deficiency while maintaining automation.
Solution Approach 2:
The patent segments the information extraction process into two distinct stages: first, the object recognizing technique extracts abstract object information (category-level), and second, the database retrieval system extracts detailed specific information (instance-level). This segmentation allows each component to specialize, with the recognition system handling automation and the database handling precision.
2Measurement precision
If manual association of reproducing time to keyword is performed, then measurement precision is improved, but productivity deteriorates
Solution Approach 1:
The system enables self-service automation where the object recognition technique automatically extracts object information from moving image content without manual intervention. The extracted information automatically queries the database to retrieve corresponding keywords and detailed information, eliminating the need for manual association while maintaining high precision through the structured database.
Solution Approach 2:
The patent implements preliminary action by pre-storing detailed object information (images, names, descriptions) in the database before the actual keyword extraction process. This pre-prepared database allows the automated system to quickly retrieve accurate information without requiring manual association during the content processing, thereby maintaining both precision and productivity.
Data Source
AI summary
An information providing device according to the exemplary embodiment includes an object recognizing unit, a retrieving unit, an obtaining unit, and a transmitting unit. The object recognizing unit extracts an image of a specific object which appears in an image of a moving image content to be distributed to a terminal device. The retrieving unit requests a retrieval device to retrieve a similar image with the image of the specific object as a retrieval key and obtains a retrieval result from the retrieval device. The obtaining unit obtains recommend information related with an image of the specific object which appears in an image of a moving image content based on the retrieval result. The transmitting unit transmits the recommend information to the terminal device.