Store Payment Tracking Using 3D Pose and Shelf Weight Triggers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing unmanned store payment systems face challenges in accurately identifying which user wants to purchase which product, particularly when using general-purpose cameras, leading to low accuracy and the need for additional verification.
Innovation Solution
An electronic device and method that utilizes a processor to receive image data from multiple sensors, generate 2D and 3D pose data for users, track their movements, identify triggers of product addition or removal, and calculate reliability to enhance user and product identification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a special camera (LiDAR or depth camera) is used to identify users and products, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent creates a virtual copy of the physical store environment by generating a 3D virtual space that mirrors the real store layout, shelves, and products. This virtual model allows the system to track user movements and product interactions without requiring complex specialized cameras, thereby maintaining measurement precision while reducing device complexity
Solution Approach 2:
The patent replaces the mechanical/optical measurement system (specialized cameras like LiDAR) with a computational approach using general-purpose cameras combined with AI algorithms. The system processes images from standard cameras to generate 2D pose data, converts it to 3D pose data, and tracks user-product interactions through software-based analysis rather than hardware-based depth sensing
2Device complexity
If a general-purpose camera (RGB camera) is used to identify users and products, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The patent transforms 2D image data from general-purpose cameras into 3D pose information through coordinate system conversion. By generating 2D pose data from multiple camera angles and converting it to 3D pose data, the system recovers depth and spatial information that would normally require specialized cameras, thereby maintaining measurement precision while using only standard cameras
Solution Approach 2:
The patent combines multiple data sources and processing methods to create a composite analytical approach: integrating 2D pose data from general-purpose cameras, converting it to 3D pose data, combining it with weight sensor data, and processing it through AI algorithms. This composite approach compensates for the limitations of individual components and achieves high measurement precision
3Device complexity
If only image data from general-purpose cameras is used, then device complexity is reduced, but reliability deteriorates due to low accuracy requiring additional verification
Solution Approach 1:
The patent merges multiple independent data sources into a unified analysis system: combining 3D pose data from general-purpose cameras with weight change data from sensors on shelves. This multi-source data fusion allows the system to cross-validate information and reliably identify user-product interactions without requiring additional verification steps, thereby improving reliability while maintaining simple device configuration
Solution Approach 2:
The patent implements a feedback mechanism where the system continuously monitors weight changes on shelves and correlates them with user 3D pose data. When a weight change is detected, the system uses the pre-tracked 3D pose information to immediately identify which user caused the change, creating a reliable closed-loop verification system that eliminates the need for manual checking
Data Source
AI summary
A electronic device and method for providing store payment service are provided. An electronic device for providing store payment service comprising a processor; and a memory operatively coupled to the processor, wherein the memory, when executed, causes the processor to identify first tracking data of the tracking data corresponding to a first time period including the trigger time point, identify a user candidate group for at least some of the at least one user based on the first tracking data, calculate a first reliability for each of the at least some users included in the user candidate group, identify that a user related to the trigger is a first user based on the first reliability, and store instructions for performing an update related to the trigger for a virtual shopping cart of the first user.