Training Image Composition for Reusable Object Detection Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning in object detection tasks faces high calculation loads and long processing times due to the need for image preprocessing, which varies across iterations, making it difficult to reuse calculation results and thus reduce processing load.
Innovation Solution
An information processing device and method that selects a base image with a target region, generates a processing target image, combines target regions from other images, and calculates features to create a dataset for machine learning, reducing the need for image preprocessing and enabling reuse of calculated features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image preprocessing is performed for each iteration in machine learning, then learning accuracy is improved, but calculation load increases and processing time lengthens
Solution Approach 1:
The patent applies preliminary action by performing image preprocessing before dataset generation, so that preprocessing is done once on the base dataset rather than repeatedly during machine learning iterations. The preprocessed images are stored and reused throughout the learning process, eliminating redundant preprocessing operations while maintaining learning accuracy.
2Measurement precision
If image preprocessing is performed repeatedly in each iteration, then learning accuracy is maintained, but processing time increases
Solution Approach 1:
The patent performs image preprocessing in advance before dataset generation, storing the preprocessed results for reuse during machine learning iterations. This preliminary action eliminates repeated preprocessing operations that would otherwise consume excessive processing time while maintaining the quality needed for accurate learning.
3Productivity
If calculation results are reused to reduce processing load, then productivity improves, but it becomes difficult to maintain learning accuracy when preprocessing varies across iterations
Solution Approach 1:
The patent performs image preprocessing before dataset generation and stores the preprocessed images for reuse. By fixing the preprocessing step prior to dataset creation, the same preprocessed images can be repeatedly used during machine learning iterations without variation, enabling calculation result reuse while maintaining consistent learning accuracy.
Data Source
AI summary
An information processing device according to the present invention performs operations including: selecting a base image from a base dataset including a target region including an object to be subjected to machine learning and a background region not including an object to be subjected to machine learning, and generating a processing target image; selecting the target region included in another image included in the base dataset; combining an image of the selected target region and information on an object to be subjected to machine learning included in the image of the target region with the processing target image; generating a dataset of the processing target images obtained by combining a predetermined number of the target regions; calculating a feature of an image included in the dataset; generating a learned model using first machine learning using the feature and the dataset; and outputting the generated learned model generated.