Vision Mesh Network for Self-Checkout Camera Sharing Across Stations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Self-checkout systems at retail locations face accuracy and reliability issues due to obstructed camera views, leading to errors in item scanning and increased 'shrinkage', as cameras on individual stations have limited access to data from other stations and may not capture all items or actions accurately.
Innovation Solution
A 'vision mesh network' connects edge cameras across multiple self-checkout stations, allowing data sharing and utilizing peripheral views from adjacent stations to enhance the quality and quantity of data inputs for computer vision modules, improving accuracy and reliability by enabling real-time processing without increasing network traffic or latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If cameras are installed on individual self-checkout stations, then each station can monitor its own area, but the camera views become obstructed and cannot capture all items or actions accurately
Solution Approach 1:
The patent merges camera resources across multiple self-checkout stations into a shared network. Cameras from adjacent stations are pooled together to create composite views of individual stations, ensuring complete coverage without obstructions while utilizing existing hardware assets.
Solution Approach 2:
The system transitions from a single-station camera perspective to a multi-station networked perspective. By combining peripheral views from neighboring stations, the system creates a comprehensive three-dimensional understanding of each checkout area, eliminating blind spots and obstructions.
2Reliability
If data is shared across multiple self-checkout stations, then accuracy and reliability improve, but network traffic and latency increase
Solution Approach 1:
The patent implements local processing at each self-checkout station where computer vision modules analyze video feeds from relevant cameras. This distributed approach processes data locally rather than centralizing all video streams, reducing network traffic while maintaining system-wide accuracy through shared camera access.
Solution Approach 2:
The system segments the video monitoring function into station-specific processing tasks. Each station processes its own checkout data using locally available camera feeds from the network, rather than one centralized system processing all feeds, which reduces overall network bandwidth requirements.
3Measurement precision
If multiple cameras are used to provide alternate views, then measurement precision improves, but device complexity increases
Solution Approach 1:
The patent makes camera assets universal across the self-checkout network. Each camera can serve multiple stations depending on positioning and operational needs, allowing a single camera to provide monitoring for multiple checkout areas at different times, reducing total camera count while maintaining precision.
Data Source
AI summary
A point-of-sale system is provided and includes a first checkout station and associated first edge cameras. The first cameras have a primary viewing area within the first checkout station and a peripheral viewing area outside the first checkout station. The system includes a second checkout station near the first checkout station and associated second edge cameras. The second cameras have a primary viewing area within the second checkout station and a peripheral viewing area outside the second checkout station. The peripheral viewing area of one second camera is within the first checkout station. The system further includes a vision mesh network having nodes in communication with each other. Some of the first and second cameras are nodes on the vision mesh network. A first edge camera receives and processes information about the first checkout station from the at least one second camera. A method is also provided.