Image-Based Ad Targeting Using Visual Recognition Engines

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current online advertising methods, such as keyword-based search and ad syndication, are limited in their ability to effectively target users with image-based content, as they do not utilize image recognition technologies to match images with relevant advertisements.

Innovation Solution

An image-based ad targeting system that receives images from mobile devices, uses optical character recognition, rigid textured object recognition, face recognition, and articulate object recognition engines to match images with stored representations, and presents relevant advertisements based on these matches.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If keyword-based search and ad syndication are used for online advertising, then advertising reach can be extended to additional partners, but the ability to effectively target users with image-based content is limited

Engineering Contradiction:

Improveadvertising reachVSAvoidimage-based targeting accuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent replaces traditional keyword-based text processing with image recognition technology. The system uses optical character recognition (OCR), rigid textured object recognition, face recognition, and articulate object recognition engines to automatically analyze and interpret visual content in images, enabling accurate image-based advertising targeting without relying on manual keyword tagging.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces multiple specialized recognition engines as intermediary components between the uploaded image and the advertising selection process. These engines (OCR, rigid textured object recognition, face recognition, articulate object recognition) act as mediators that transform raw image data into structured information that can be matched with advertising content, thereby improving targeting precision while maintaining system scalability.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If multiple recognition engines are used to match images with stored representations, then object identification accuracy is improved, but system complexity increases

Engineering Contradiction:

Improveobject identification accuracyVSAvoidsystem complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the image recognition task into multiple specialized segments, each handled by a dedicated recognition engine. Instead of using one complex general-purpose recognition system, the patent segments the problem into OCR for text, rigid textured object recognition for static objects, face recognition for human faces, and articulate object recognition for objects with movable parts. This segmentation improves accuracy for each specific task while making the overall system more manageable through modular architecture.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS8315423B1Providing information in an image-based information retrieval system

Publication Date: 2012.11.20 GOOGLE LLC

AI summary

Techniques are described for providing information in an image-based information retrieval system. An image including an object is received from a mobile device over a network of computer. The object included the image is matched with a stored representation of the object. Information related to the object is identified based on an association between the identified information and the stored representation of the object. Presentation, over the network of computers on the mobile device, of the identified information is enabled.