Pressure-Sensitive Apparel Image Search via CNN Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users face difficulties in describing and searching for clothing items in images using textual descriptions, as fashion images are complex and require precise identification of objects within the images.
Innovation Solution
A touch and pressure-based image searching method that uses convolutional neural networks (CNNs) to detect user input on a screen, quantify pressure, and retrieve similar products from e-commerce websites, allowing users to select objects of interest without segmenting the entire image, and retrieve corresponding products based on pressure-sensitive interactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If users use textual descriptions to search for clothing items, then the search process becomes complex and imprecise, but the system requires detailed object identification within images
Solution Approach 1:
The patent replaces the mechanical/manual process of textual description and keyword searching with an automated visual recognition system using CNNs. The system automatically identifies and segments clothing objects in images through deep learning, eliminating the need for users to manually describe items textually while achieving precise object identification through automated image analysis and feature extraction
Solution Approach 2:
The patent introduces an intermediary visual search system that acts as a mediator between the user's simple image selection and the complex product database. The CNN-based object detection and image segmentation algorithms serve as intermediaries that automatically analyze image content, identify clothing items, extract features, and match them with product catalogs, bridging the gap between simple user input and precise product retrieval
2Measurement precision
If users manually segment and describe each clothing item in an image, then object identification becomes precise, but the operation complexity and time required increase significantly
Solution Approach 1:
The patent implements self-service automation where the system performs object detection, image segmentation, and feature extraction automatically without requiring user intervention. The CNN-based algorithms autonomously analyze the uploaded image, identify multiple clothing items, segment them from the background, extract relevant visual features, and generate search queries automatically, making the system self-sufficient in performing complex image analysis tasks
Solution Approach 2:
The patent performs preliminary actions by pre-processing images through automated object detection and segmentation before the actual product search. The system pre-extracts visual features from clothing items in the uploaded image, pre-segments relevant objects from complex backgrounds, and pre-generates search queries based on identified items, so that when the user submits the image, the heavy computational work has already been completed in advance
3Measurement precision
If the system retrieves only the touched object, then search precision is high, but the user may miss related products within the same image region
Solution Approach 1:
The patent implements dynamic search scope adjustment based on user interaction. When a user touches or selects a region in the uploaded image, the system dynamically expands the search scope from the single touched object to include all detected clothing items within that selected region. The search parameters and object list are dynamically adjusted based on the spatial extent of the user's selection, allowing flexible expansion or contraction of the search area to match user intent
Data Source
AI summary
Methods, systems, and computer program products for pressure-based apparel image searching are provided herein. A computer-implemented method includes converting images in a product catalog of an electronic commerce website to a predetermined representation; storing the converted images in an index; determining a first object of interest within an image derived from a social media post and displayed on a screen, by detecting physical contact imparted by a user at a position on the screen corresponding to where the first object of interest is located; quantifying the amount of pressure applied by the user via the physical contact; determining additional objects of interest within the image based on the amount of pressure applied by the user; retrieving, from the index, images of products corresponding to the first object of interest and images of products corresponding to the additional objects of interest; and displaying the retrieved images on the screen.