Visual Clothing Retrieval via Image Segmentation and Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining relevant advertisements based on image content are inefficient due to the need for human labeling, which is time and labor-intensive and does not scale well with increasing image volumes.
Innovation Solution
The approach involves segmenting and clustering image segments in a query image to identify product items, generating an articulated pose estimation and product probability map, and classifying these segments to search for visually similar product images in a database, allowing for efficient retrieval of relevant product images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If human labelers are used to manually identify elements in images and select advertisements, then advertisement relevance to image content is improved, but time consumption and labor intensity increase significantly
Solution Approach 1:
The patent replaces the mechanical human labeling system with an automated computer vision system that uses image processing algorithms to identify elements in images and retrieve relevant advertisements. The system automatically analyzes image content, extracts features, and queries the advertisement database without human intervention, thereby eliminating time consumption and labor intensity while maintaining advertisement relevance.
Solution Approach 2:
The system enables self-service by allowing the computer to autonomously perform the entire workflow from image analysis to advertisement selection. The automated system processes images, identifies elements, and retrieves relevant advertisements independently, freeing human operators from manual labeling tasks while ensuring consistent and scalable advertisement relevance.
2Measurement precision
If human labelers are used to manually identify elements in images, then accurate image content identification is achieved, but labor intensity and processing cost increase
Solution Approach 1:
The patent replaces complex human cognitive processes with automated computer vision algorithms that systematically analyze image content. The system uses feature extraction, template matching, and database querying to identify elements and retrieve advertisements, achieving accurate identification while reducing system complexity through standardized automated processes.
Solution Approach 2:
The patent segments the image processing task into distinct automated stages: feature extraction, element identification, and advertisement retrieval. By dividing the complex task into manageable automated components, the system achieves accurate image content identification while keeping each processing stage relatively simple and maintainable.
3Reliability
If manual advertisement selection based on human labeling is used, then relevant advertisements can be displayed, but the process does not scale efficiently with increasing image volumes
Solution Approach 1:
The patent replaces manual advertisement selection with an automated system that processes images and retrieves advertisements programmatically. This substitution enables the system to handle increasing image volumes efficiently, as the automated process can be parallelized and scaled horizontally without proportionally increasing human resources, thereby maintaining advertisement relevance while improving productivity.
Solution Approach 2:
The patent creates a universal automated system that can handle diverse image types and advertisement queries through a single integrated platform. The system's multi-functional design allows it to process various image formats, extract different types of features, and retrieve relevant advertisements from a centralized database, enabling efficient scaling across different applications and image volumes.
Data Source
AI summary
Techniques are provided for efficiently identifying relevant product images based on product items detected in a query image. In general, a query image may represent a digital image in any format that depicts a human body and one or more product items. For example, a query image may be an image for display on a webpage, an image captured by a user using a camera device, or an image that is part of a media content item, such as a frame from a video. Product items may be detected in a query image by segmenting the query image into a plurality of image segments and clustering one or more of the plurality image segments into one or more image segment clusters. The resulting image segments and image segment clusters may be used to search for visually similar product images.