Data Creation Apparatus Using Accessory Metadata for Training Data Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Selecting appropriate image data for machine learning training data from vast datasets is labor-intensive and costly due to the need for manual annotation, which increases the cost of creating training data.
Innovation Solution
A data creation apparatus that utilizes accessory information recorded with image data to efficiently select and create training data by setting conditions based on permission, history, purpose, and creator information, and suggests additional conditions to enhance the dataset.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual selection and annotation of image data is performed, then the quality and appropriateness of training data is improved, but the time and cost required for data creation increases significantly
Solution Approach 1:
The system enables image data to select itself by using accessory information automatically recorded during image capture (such as location, time, device metadata) to autonomously determine suitability for training, eliminating the need for manual human selection and annotation while maintaining data quality
Solution Approach 2:
Accessory information is recorded and prepared in advance during the image capture process, so that when training data selection is needed, the pre-recorded metadata can be immediately utilized for automated filtering and selection without requiring subsequent manual processing
2Reliability
If manual selection and annotation of image data is performed, then the quality and appropriateness of training data is improved, but the cost of data creation increases
Solution Approach 1:
The system enables image data to select itself by using accessory information automatically recorded during image capture (such as location, time, device metadata) to autonomously determine suitability for training, eliminating the need for manual human selection and annotation while maintaining data quality
Solution Approach 2:
The manual mechanical process of human annotators reviewing and selecting images is replaced with an automated information processing system that uses accessory metadata to filter and select appropriate training data, significantly reducing labor costs
3Productivity
If accessory information is utilized for automated selection, then the speed of data creation is improved, but the complexity of the selection system increases
Solution Approach 1:
The accessory information serves multiple functions simultaneously: it acts as both metadata for the image and as selection criteria for training data filtering, allowing a single information structure to support both image management and machine learning training requirements
Solution Approach 2:
Accessory information acts as an intermediary layer between the raw image data and the training data selection process, providing structured metadata that bridges the gap between image capture and machine learning requirements without requiring direct complex processing of the images themselves
Data Source
AI summary
Image data to be used for creating training data is appropriately selected from a plurality of pieces of image data using information included in accessory information recorded in the image data.One embodiment of the present invention is a data creation apparatus that creates training data used in machine learning from image data in which accessory information is recorded, the data creation apparatus being configured to execute setting processing of setting any setting condition related to first information related to the machine learning or to second information related to a creator of the image data, a creator of the accessory information, or a right holder of the image data included in the accessory information of the image data with respect to a plurality of pieces of the image data in which the accessory information including the first information or the second information is recorded, and creation processing of creating the training data based on selection image data in which the first information or the second information satisfying the setting condition is recorded.