Merchandise specification systems and programs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing merchandise specification systems face challenges in accurately identifying displayed merchandise due to varying lighting conditions and high processing loads, especially when comparing image recognition results with sample images, leading to low accuracy and increased resource requirements.
Innovation Solution
A merchandise specification system that performs keystone correction on photographed images to generate front-facing image information, automatically specifying reference vertices and reducing the processing load by comparing image features within a predetermined range, thereby improving accuracy and efficiency in merchandise identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If image recognition techniques are used to automatically specify merchandise from photographed images, then manual burden is reduced, but recognition accuracy decreases due to varying lighting conditions
Solution Approach 1:
The system performs preliminary actions by creating multiple candidate specifications through different image recognition methods before final determination. It generates candidate merchandise specifications using both sample-image-based recognition and adjacent-face-based recognition, then selects the most reliable candidate, thereby improving accuracy while maintaining automation.
Solution Approach 2:
The system uses feedback from multiple recognition results to improve accuracy. It compares candidate specifications from different methods, checks consistency with adjacent faces, and uses this feedback to determine the final merchandise specification, resolving the accuracy issue while keeping the system automated.
2Measurement precision
If image recognition based on sample images is performed, then merchandise can be identified, but processing load and resource requirements increase significantly
Solution Approach 1:
The system applies partial action by using adjacent-face-based recognition as the primary method, which requires less processing power. It only uses sample-image-based recognition partially, mainly for reference or when adjacent faces are unavailable, thereby reducing overall processing load while maintaining identification capability.
Solution Approach 2:
The system segments the recognition process into multiple stages: first generating candidates from adjacent faces (low processing load), then selectively using sample image comparison (higher processing load) only when needed. This segmentation reduces the overall processing burden while preserving identification accuracy.
3Reliability
If multiple image recognition methods are used to improve accuracy, then merchandise specification reliability increases, but processing time and computational resources increase
Solution Approach 1:
The system dynamically adjusts the recognition process based on available information. It first attempts the faster adjacent-face-based method, and only invokes the more time-consuming sample-image-based method when necessary (e.g., when adjacent faces are unavailable or results are ambiguous), thereby optimizing the balance between reliability and processing time.
Solution Approach 2:
The system performs preliminary filtering using the faster recognition method before applying the more resource-intensive method. By pre-processing with adjacent-face-based recognition and only proceeding to sample-image-based recognition when needed, it reduces overall processing time while maintaining high specification reliability.
Data Source
AI summary
A merchandise specification system has: a keystone correction processing unit that performs a keystone correction process on photographed image information obtained by photographing the display shelf to generate front-facing image information; and a merchandise identification information specification processing unit that specifies merchandise on a face in the front-facing image information.