Image Content Based Advertisement Targeting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current internet-based video streaming and image browsing advertisements are often tangentially related to the content, leading to a lower conversion rate due to their lack of relevance.
Innovation Solution
A system that enables users to select areas of images or videos, analyzing and embedding related advertisements or information, allowing for targeted advertising by associating text or objects within the content with relevant ads, and displaying these ads in a more relevant manner.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If advertisements are inserted between scenes of videos or adjacent to images, then advertising space can be provided and revenue can be generated, but the advertisements become tangentially related to the content, reducing effectiveness and conversion rate
Solution Approach 1:
The patent segments the advertisement delivery system into multiple components: image analysis module that extracts features from video frames, advertisement matching module that compares extracted features with advertisement databases, and selective insertion module that places ads only when relevance thresholds are met. This segmentation enables targeted advertising while maintaining content coherence.
Solution Approach 2:
The system implements feedback mechanisms by analyzing user interactions with advertisements and content, using this data to refine future advertisement selections. The image analysis module continuously learns from viewer behavior patterns to improve matching accuracy, ensuring advertisements remain relevant while maximizing revenue potential.
2Adaptability or versatility
If advertisements are targeted based on search terms or general website content, then some relevance can be achieved, but the relevance remains tangential rather than directly connected to specific visual content
Solution Approach 1:
The patent replaces traditional text-based keyword matching mechanisms with image-based feature extraction and recognition systems. By analyzing visual elements directly from video frames using computer vision algorithms, the system achieves precise matching between advertisement content and visual scene elements, transcending the limitations of text-only search term targeting.
Solution Approach 2:
The system changes the fundamental parameters of advertisement targeting from textual metadata (search terms, website categories) to visual feature parameters extracted directly from image content. This includes color histograms, object detection results, scene recognition labels, and spatial relationship features, enabling direct visual-content based advertisement matching.
Data Source
AI summary
A system for serving an advertisement in a networked environment receives data that defines a user selection of an image. The system also selects an advertisement associated with the user selection and communicates the selected advertisement to the user. The data that defines the user selection includes an x selection coordinate, a y selection coordinate, a width, and a height that defines a region of an image. The user selection defines also defines a scene of a video. The system also includes circuitry and logic configured to extract text and to detect objects from an image region defined by the user selection and to select an advertisement associated with any extracted text and/or detected object.