Product Identification Model via Image Segmentation and Selective Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional product identification methods face challenges in accurately identifying products with irregular shapes, leading to decreased accuracy due to difficulties in extracting feature quantities from mixed and non-fixed shape products.
Innovation Solution
A learned model generating method that divides product images into areas, extracts specific images based on product presence conditions, and performs machine learning to generate an identifying model, focusing on images with high product occupancy rates and excluding non-product elements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional feature extraction methods are used for product identification, then the process is simple, but the identification accuracy decreases for products with irregular shapes
Solution Approach 1:
The patent divides the product image into multiple divided images (areas), extracting feature quantities from each region separately. This segmentation approach allows the system to handle irregularly shaped products by processing manageable sections, thereby improving identification accuracy without overwhelming computational complexity
Solution Approach 2:
The patent extracts only the necessary divided images that contain actual product information, filtering out areas with containers or backgrounds. This selective extraction focuses computational resources on relevant regions, improving both accuracy and efficiency by eliminating unnecessary processing
2Measurement precision
If all divided images are used for machine learning, then more data is available for training, but the processing load increases and includes irrelevant non-product elements
Solution Approach 1:
The patent extracts only divided images that satisfy predetermined conditions (containing actual product, not just container or background). This selective extraction provides high-quality training data for machine learning while reducing processing load by excluding irrelevant images, thus improving both model accuracy and processing efficiency
Solution Approach 2:
The patent applies different processing approaches to different divided images based on their content characteristics. Images containing product are processed with full attention for training, while images containing only containers or backgrounds are filtered out or processed differently, optimizing resource allocation according to local image quality
Data Source
AI summary
A measuring system 1 includes a server 200 identifying a kind of a product from a product image in which the product is included and a measuring device 100 identifying the kind of the product from the target image in which the product is included. The server 200 includes an acquisition unit that acquires a product image and product information relating to a kind of a product, a dividing unit that acquires a plurality of divided imaged by dividing the product image into a plurality of areas, and a generation unit that generates an identifying model by performing machine learning on the basis of a plurality of divided images extracted by an extraction unit that extracts a plurality of divided images satisfying a predetermined condition relating to a shown amount of the product from among the plurality of divided images.