Congestion Prediction Using Image Data Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for predicting future congestion degrees in areas, such as those based on past positioning information, face increased processing loads and complex data management, especially when dealing with observation values associated with spatial coordinates.
Innovation Solution
An information processing device uses machine learning to generate predicted image data by transforming actual image data containing current congestion degrees into future congestion degree data, using a deep learning model like PredNet, which reduces processing load and simplifies data management by focusing on spatial patterns without requiring detailed feature analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional methods are used to calculate and cluster congestion degrees for each spot or area, then future congestion prediction can be achieved, but processing load increases and data management becomes complicated
Solution Approach 1:
The patent creates a simplified copy of the spatial data structure by representing congestion degrees as image data where spatial coordinates are mapped to pixel positions. This copy retains the essential spatial patterns while eliminating the complexity of managing detailed location information, making data processing more efficient while preserving prediction accuracy
Solution Approach 2:
The patent transforms the parameter representation of congestion data by converting numerical congestion degrees into visual image data parameters (pixel values). This parameter transformation simplifies data management operations while maintaining the ability to perform accurate congestion prediction through image processing techniques
2Measurement precision
If detailed spatial features are analyzed for congestion prediction, then prediction accuracy improves, but computational burden increases
Solution Approach 1:
The patent extracts only the essential spatial pattern information from location data by mapping congestion degrees to image pixel values. This extraction process removes unnecessary detailed spatial features while retaining the core information needed for accurate congestion prediction, thereby reducing computational requirements
Solution Approach 2:
The patent replaces complex mechanical computation of spatial relationships with automated image processing operations. By treating congestion data as images, the system uses efficient image analysis algorithms instead of computationally intensive spatial calculations, reducing power consumption while maintaining prediction accuracy
Data Source
AI summary
An information processing device includes a communication unit that acquires first image data in which an observation value observed at a time t is used as a pixel value and a learning processing unit that generates second image data in which an observation value predicted to be observed at a time t+n after the time t is used as a pixel value from the first image data acquired by the acquiring unit based on a learning model obtained by machine learning using the first image data, in which the machine learning occurs based on a comparison of the first image data in which an observation value observed at a target time is used as a pixel value and the second image data in which an observation value predicted to be observed at the target time is used as a pixel value.