Shopping Cart Scale Estimation Using 2D-to-3D Skeleton Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Fraud detection systems using surveillance cameras struggle to accurately estimate the scale of a person relative to a shopping cart due to the two-dimensional nature of captured images, leading to incorrect behavior recognition.
Innovation Solution
An estimation program that uses existing object and skeleton detection technologies to generate three-dimensional skeleton information, estimate the scale of a person based on the grip part of a cart, and apply homography transformation to determine absolute three-dimensional coordinates, enabling accurate behavior recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If two-dimensional surveillance camera images are used to detect customer behavior, then the system can monitor the store, but the scale of the person relative to the cart cannot be accurately estimated
Solution Approach 1:
The patent transforms two-dimensional image data into three-dimensional spatial information by detecting the cart's grip part position and using skeleton information to calculate the person's scale. This dimensionality change enables accurate scale estimation by inferring depth relationships from 2D images through geometric calculations based on the cart's known dimensions and the person's skeleton joints.
2Reliability
If the scale of the person is not accurately estimated, then the system is simple, but behavior recognition becomes incorrect
Solution Approach 1:
The patent uses the cart's grip part as an intermediary reference object to estimate the person's scale. By detecting the grip part's position in the image and using it as a reference point with known dimensions, the system can calculate the person's scale through geometric relationships, thereby improving behavior recognition accuracy without requiring complex 3D sensing equipment.
3Measurement precision
If three-dimensional skeleton information is generated from two-dimensional images, then scale estimation accuracy improves, but processing complexity increases
Solution Approach 1:
The patent replaces complex mechanical 3D sensing systems with computational methods that process two-dimensional image data. By using skeleton detection algorithms and geometric calculations based on the cart's grip part position, the system achieves 3D scale estimation through software processing rather than hardware complexity, thereby reducing device complexity while maintaining measurement precision.
Data Source
AI summary
A non-transitory computer-readable recording medium stores therein an estimation program that causes a computer to execute a process including, identifying a first person who uses a first cart from a first image acquired by capturing inside a store, generating skeleton information of the first person, acquiring, by using the skeleton information, a first space in which the first person grasps a grip part of the first cart, and estimating a first scale of the first person based on the first space and length information of the first cart.