Disposable Income Estimation via Image Object Frequency Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems struggle to provide accurate recommendation information based on image analysis, as they may incorrectly detect subjects like vehicles and lack specific weighting coefficients for user asset estimation, leading to inappropriate recommendations.
Innovation Solution
An image processing system that analyzes a group of images to recognize objects, derives disposable income ranges, and applies weighting coefficients based on appearance frequency to estimate user disposable income, thereby providing tailored recommendation information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single image is analyzed to detect subjects, then the processing speed is fast, but the accuracy of user asset estimation deteriorates due to inability to distinguish between user-owned objects and accidentally captured objects
Solution Approach 1:
The patent segments the image analysis process into multiple independent components: object detection, frequency analysis across multiple images, and weighted scoring. By dividing the analysis across multiple images rather than relying on a single image, the system can distinguish between objects that appear frequently (likely user-owned) and those that appear occasionally (likely accidentally captured), thereby improving estimation accuracy without requiring overly complex single-image analysis
Solution Approach 2:
The system performs preliminary analysis by collecting and analyzing multiple images before making the final asset estimation. By gathering frequency data from multiple images in advance and calculating weights based on appearance patterns, the system prepares comprehensive information before generating the final estimation, improving accuracy while keeping the actual decision-making process relatively simple
2Measurement precision
If weighting coefficients are applied based on appearance frequency, then the recommendation accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent changes the parameter from simple object detection to frequency-based weighted scoring. By analyzing how often objects appear across multiple images and assigning weights based on these frequencies, the system transforms basic detection data into more meaningful asset estimation metrics. This parameter change improves recommendation accuracy by considering temporal patterns rather than just single-instance detections, while the weighting mechanism remains computationally manageable through systematic calculation
3Reliability
If multiple images are analyzed to improve estimation accuracy, then the reliability of disposable income estimation improves, but the processing time increases
Solution Approach 1:
The patent segments the image group into multiple analysis target images and processes them systematically. By dividing the overall analysis task into manageable segments (individual image analyses followed by aggregation), the system can improve reliability through multiple data points while avoiding the need to process all images simultaneously, thus reducing overall processing time through efficient resource utilization
Solution Approach 2:
The system performs analysis on a representative subset of images from the image group rather than requiring exhaustive analysis of every single image. By selecting and analyzing multiple key images that provide sufficient frequency information, the system achieves reliable estimation without the excessive time cost of analyzing every possible image, balancing reliability with processing efficiency
Data Source
AI summary
Provided are an image processing system, an image processing method, and a program that can provide recommendation information according to a disposable income estimated on the basis of an analysis result of an image group.The image processing system, the image processing method, and the program include recognizing an object of an analysis target image of a user (44), converting information on the object into disposable income range information of the user (50), acquiring accessory information including imaging date information of the analysis target image, and deriving an appearance frequency of the object on the basis of the imaging date information (52), deriving a weighting coefficient on the basis of the appearance frequency information of the object (54), estimating the disposable income of the user using the disposable income range information and the weighting coefficient (56), and transmitting recommendation information corresponding to the disposable income of the user (58).