Sketch-Based Content Search Using AI Image Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for searching content on devices are inefficient and inconvenient, as they require keyword-based searches or manual tagging of metadata, making it difficult to find images based on atmosphere or color.
Innovation Solution
An electronic device and control method that uses a sketch drawn by a user to generate machine-based images through an AI model, allowing for content search based on visual elements like color and shape, with the AI model trained using generative adversarial networks or variational auto-encoders.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If keyword-based search or manual metadata tagging is used, then search functionality is provided, but search efficiency and convenience deteriorate
Solution Approach 1:
The patent replaces manual keyword-based search and metadata tagging with an AI-based image generation and recognition system. Users draw sketches instead of typing keywords, and the AI model automatically generates images and performs semantic understanding, eliminating the need for manual metadata creation and improving both search efficiency and convenience.
Solution Approach 2:
The patent introduces an AI model as an intermediary between the user's sketch and the content database. The AI model generates images from sketches, performs semantic understanding, and retrieves relevant content, acting as a smart mediator that translates simple user input into accurate search results without requiring users to manually create complex search queries.
2Adaptability or versatility
If keyword search is used, then basic search functionality is achieved, but the ability to search by atmosphere or color deteriorates
Solution Approach 1:
The patent changes the search parameter from text-based keywords to visual-based sketches. By allowing users to draw images and using AI to generate and analyze visual content, the system can search by atmosphere, color, and visual characteristics that cannot be captured by traditional keyword search, thereby improving search versatility and preserving visual information.
Solution Approach 2:
The patent adds a visual dimension to the search process by introducing image generation and recognition capabilities. Instead of searching only through text keywords, users can search through visual sketches, and the AI model processes images to retrieve content based on visual attributes like atmosphere and color, expanding the search into a new dimensional space.
3Ease of manufacture
If manual metadata tagging is used, then content organization is achieved, but user convenience and time consumption worsen
Solution Approach 1:
The patent implements self-service by enabling the AI model to automatically generate metadata and organize content without requiring manual user input. The system autonomously performs image generation, semantic understanding, and content classification based on user sketches, eliminating the time-consuming manual tagging process while maintaining effective content organization.
Solution Approach 2:
The patent performs preliminary action by having the AI model pre-process and organize content in advance. When users submit sketches, the AI model proactively generates images, creates metadata, and organizes content before the user needs to search, thereby eliminating the need for users to spend time on manual tagging and content organization tasks.
Data Source
AI summary
A method for controlling an electronic device is disclosed. The method for controlling an electronic device comprises the steps of: receiving a sketch drawn by a user; acquiring at least one machine-generated image on the basis of the sketch by using an artificial intelligence model learned by an artificial intelligence algorithm; displaying the acquired at least one machine-generated image; receiving a user command for selecting one of at least one displayed machine-generated image, and searching for at least one piece of content corresponding to the image selected according to the user command; and providing the searched for at least one piece of content.