Machine-Learning Item Pricing From Condition Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems fail to automatically adjust selling prices based on the condition of items such as discoloration or deformation, which can affect the value of perishable goods like fresh foods.
Innovation Solution
A system comprising a store server and a portable terminal with a camera, display, and communication controller, which uses machine-learning models to recognize item conditions and set dynamic selling prices accordingly, superimposing the prices onto captured images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional barcode scanning systems are used, then item identification is simple, but the system cannot automatically adjust prices based on item condition such as discoloration or deformation
Solution Approach 1:
The patent replaces manual inspection and fixed pricing systems with an automated image recognition system using deep learning models. The machine learning model automatically analyzes item conditions (discoloration, deformation) from captured images and dynamically adjusts prices without human intervention, resolving the contradiction between adaptability and complexity by using intelligent automation rather than complex mechanical systems
Solution Approach 2:
The system changes the pricing parameter dynamically based on detected item conditions. Instead of using fixed prices, the system adjusts the selling price parameter according to the degree of discoloration or deformation detected by the machine learning model, enabling adaptive pricing that reflects actual item quality while maintaining system simplicity through automated parameter adjustment
2Productivity
If manual inspection of item conditions is performed, then pricing accuracy reflects item quality, but store efficiency decreases due to manual labor requirements
Solution Approach 1:
The system enables self-service condition detection where the machine learning model automatically inspects items and determines pricing without requiring store employees to manually check each item. The automated image recognition system performs the inspection function itself, significantly improving store efficiency while maintaining high detection accuracy through sophisticated algorithms
Solution Approach 2:
The patent substitutes manual mechanical inspection with automated digital image recognition. Instead of human eyes and hands physically examining items, the system uses camera-based image capture and machine learning algorithms to detect conditions, thereby eliminating manual labor while achieving consistent and accurate condition assessment
3Loss of information
If fixed selling prices are used for all items, then operation is simple, but customer transparency about item condition value is reduced
Solution Approach 1:
The system implements feedback by continuously capturing images of items, analyzing their conditions through machine learning, and adjusting prices accordingly. This closed-loop feedback mechanism ensures that pricing information accurately reflects item conditions, providing customers with transparent pricing while the automated feedback process maintains operational simplicity
Solution Approach 2:
The machine learning model acts as an intermediary between the physical item condition and the pricing information. It processes visual data from images and translates it into appropriate price adjustments, serving as an intelligent mediator that bridges the gap between item quality assessment and pricing operation, thereby providing transparency without complicating the user experience
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system for setting and displaying a selling price of an item, includes a server that stores item information; and a terminal including: a camera, a display, a communication controller, and a processor configured to: capture an image, display the image, and transmit the image to the server, and upon receipt of a selling price of an item from the server, control the display to display the price over the image. The store server is configured to, upon receipt of an image from the terminal: execute image recognition on the image and identify an item, input the image to a machine-learning model trained to recognize a discoloration state or a deformation state of the item in the image, set a selling price for the item based on the discoloration state or the deformation state, and transmit the price to the terminal.