Bakery Tray Validation Using Multi-Angle Imaging and Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In distribution centers, efficiently validating and restacking mixed trays of different bakery products to fulfill orders accurately is challenging due to the need for precise identification and quantity verification of various bakery products.
Innovation Solution
A computer system that uses cameras to capture images of bakery trays, employing machine learning models to identify the type of products in each tray and compare them to orders, providing real-time confirmation or error indication to workers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual identification and verification of bakery products in trays is performed, then workers can directly verify product types and quantities, but the process is time-consuming and prone to human error
Solution Approach 1:
The patent replaces the manual mechanical inspection process with an automated optical inspection system using cameras and machine learning algorithms. The system captures images of bakery trays and automatically identifies product types, quantities, and tray configurations, eliminating the need for manual visual inspection while maintaining high accuracy and reducing validation time
Solution Approach 2:
The system creates digital copies (images) of the physical bakery trays and processes these copies through machine learning models. This allows simultaneous analysis of multiple trays without physical handling, enabling rapid validation while preserving product identification accuracy through detailed image analysis
2Adaptability or versatility
If multiple types of bakery products are mixed in single trays for restacking, then space utilization improves, but the complexity of identifying and verifying product types increases
Solution Approach 1:
The machine learning model segments the image analysis task into distinct components: detecting tray boundaries, identifying individual products, classifying product types, and counting quantities. This segmentation allows the system to handle mixed product trays systematically by processing each element separately and aggregating results
Solution Approach 2:
The patent develops a universal machine learning model that can identify multiple types of bakery products (buns, rolls, bread, pastries) and various tray configurations using a single system. The model is trained on diverse datasets and can adapt to different product types and tray arrangements, eliminating the need for separate identification systems for each product type
3Measurement precision
If automated imaging systems are used to capture trays from multiple angles, then complete product identification is achieved, but the system complexity and imaging time increase
Solution Approach 1:
The system transitions from single-angle 2D imaging to multi-angle 3D imaging by capturing trays from front, rear, and overhead perspectives. This dimensional approach allows the machine learning model to identify products regardless of their orientation or position in the tray, achieving comprehensive detection while managing complexity through coordinated multi-camera systems
Data Source
AI summary
A computer system for validating trays loaded with products receives at least one image of products in at least one tray of a plurality of trays in an imaging area. A type of the products in the at least one tray is determined based upon the at least one image. The type of products in the at least one tray is compared to at least one order. A confirmation or error is indicated based upon the comparison. In some examples, the products are bakery products of different types loaded in plastic, stackable bakery trays. In one example method, a plurality of stacks of loaded bakery trays are imaged from a front or rear of the trays. In another example method, the plurality of stacks are imaged from overhead as the trays are loaded onto the stacks.