Virtual Fitting System with Pose-Based Image Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing virtual clothes fitting technologies are time-consuming and inefficient, as they require repeated try-ons and lack the ability to effectively compare visual differences among multiple clothing sets, resulting in an unsatisfactory user experience.
Innovation Solution
A display system equipped with an image capturing device, screen, and processor that captures images of a user, segments and processes the foreground and background, and generates preview images to display on the screen, allowing for real-time comparison of clothing options in different poses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If virtual clothes fitting is implemented using basic image capture and overlay, then the fitting function is provided, but the visual realism and match with user figure are poor
Solution Approach 1:
The patent replaces traditional mechanical/physical try-on methods with an automated image processing system. The system uses pose detection algorithms and automatic image composition to substitute the manual process of repeatedly trying on clothes, achieving realistic visual effects through computational methods rather than physical interaction.
Solution Approach 2:
The patent creates accurate digital copies of the user's pose and figure from captured images. By detecting key pose points and generating standardized preview images that copy the user's actual posture, the system achieves realistic visualization without requiring physical clothes or complex 3D modeling.
2Measurement precision
If multiple clothes sets are tried on repeatedly to compare visual differences, then selection accuracy is improved, but time consumption increases significantly
Solution Approach 1:
The patent performs preliminary action by automatically generating and displaying multiple clothing preview images simultaneously before the user makes a selection. Instead of requiring sequential try-ons, the system pre-processes multiple clothing options and presents them all at once, allowing immediate comparison and reducing decision time.
Solution Approach 2:
The patent transitions from single-image sequential viewing to multi-image simultaneous display, adding the dimension of parallel comparison. By showing multiple clothing preview images together on the screen with standardized poses, the system enables users to compare options across different dimensions (styles, colors, fits) without temporal sequencing.
3Ease of operation
If the user needs to remember visual differences after repeated fitting, then no comparison function is needed, but user experience deteriorates
Solution Approach 1:
The patent implements feedback by continuously displaying the user's pose and multiple clothing preview images on the screen throughout the selection process. This visual feedback loop allows users to maintain awareness of all options without relying on memory, enabling informed decision-making through persistent visual information rather than transient recall.
Data Source
AI summary
A clothes and accessories fitting method, a display system and a computer-readable recording medium thereof are provided, where the method is applicable to a display system having an image capturing device and a screen and includes the following steps. Images of a user wearing a first apparel appearing in front of the screen are continuously or intermittently captured by using the image capturing device. When a first image of the user in a particular pose is determined to be included in the images, the first image is shot by using the image capturing device. A foreground region and a background region in the first image are segmented, and the background region is processed to generate a processed background region. A first preview image is generated according to the foreground region and the processed background region and displayed on the screen.