Virtual Shoe Try-On Using Unified Neural Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge of purchasing shoes online is that users cannot test the fit before buying, leading to difficulties in finding a suitable size and style.
Innovation Solution
A virtual shoe try-on feature using machine learning to predict how a shoe will fit on a user's foot by superimposing a 2D model of the shoe onto an image of the foot, trained with a single neural network that extracts features for various predictions, reducing computing resource consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple machine learning models are used to make different predictions for shoe fit, then prediction accuracy is improved, but computing resource consumption increases
Solution Approach 1:
The patent combines multiple separate machine learning models into a single unified model that performs all prediction tasks (foot boundary detection, shoe model selection, size recommendation) simultaneously. This integration reduces the overall computing resource consumption while maintaining the accuracy benefits of multiple specialized models.
Solution Approach 2:
The unified machine learning model is designed to perform multiple functions: detecting foot boundaries, selecting appropriate shoe models, and recommending sizes. This multi-functional approach eliminates the need for separate specialized models, reducing computational overhead while preserving prediction accuracy.
2Use of energy by moving object
If a single machine learning model is used to make all predictions, then computing resource consumption is reduced, but prediction accuracy may deteriorate
Solution Approach 1:
The patent merges multiple prediction functions into a single machine learning model that processes foot images and generates comprehensive shoe fit recommendations. This unified approach maintains accuracy by integrating the capabilities of what would otherwise be separate specialized models into one efficient system.
Solution Approach 2:
The single machine learning model is structured with multiple processing stages or modules that handle different prediction tasks sequentially or in parallel. This internal segmentation allows the model to maintain specialized functionality for each task while operating as a unified system with reduced computational overhead.
3Measurement precision
If high computing resources are allocated for virtual try-on processing, then fit prediction accuracy is improved, but device compatibility decreases
Solution Approach 1:
The patent consolidates multiple heavy computational models into a single optimized model that can run efficiently on mobile devices with limited resources. This merging enables accurate fit predictions while maintaining compatibility with consumer smartphones and tablets.
Solution Approach 2:
The machine learning model is optimized with adjusted parameters and architecture suitable for mobile deployment, balancing accuracy requirements with the computational constraints of consumer devices. This allows the system to deliver accurate fit predictions on devices with limited processing power and memory.
Data Source
AI summary
A method includes extracting, using a backbone of a machine learning model, a plurality of features from an image of a foot and predicting, using a first portion of the machine learning model and based on one or more features of the plurality of features, a first aspect of the foot. The method also includes predicting, using a second portion of the machine learning model and based on one or more features of the plurality of features, a second aspect of the foot different from the first aspect, generating, using at least the first aspect and the second aspect, a two-dimensional model of a shoe, and superimposing the two-dimensional model of the shoe onto the image of the foot.