Image Data Augmentation Using Distortion Functions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object detection algorithms face challenges in increasing the diversity of neural network augmented data due to the time-consuming and inefficient nature of manual annotation processes.
Innovation Solution
An image data augmentation apparatus and method that utilize distortion operation functions to twist pixels and convert object information in images, generating augmented images that are fed into a machine learning module for enhanced training data diversity without requiring additional manpower or time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual annotation is used to gather training data, then object detection accuracy can be improved, but the time consumption and inefficiency increase significantly
Solution Approach 1:
The patent uses pre-stored distortion operation functions as templates to generate augmented images automatically, eliminating the need for manual annotation of each image. The system copies and applies these distortion functions to transform original images and their corresponding object information, significantly reducing time consumption while maintaining data quality for training neural networks.
2Adaptability or versatility
If more training data is gathered manually, then the diversity of neural network augmented data increases, but the cost of manpower and time increases
Solution Approach 1:
The patent applies different distortion operation functions with varying parameters (such as different distortion degrees, types, and configurations) to the original images automatically. This transforms the images into diverse augmented versions while systematically converting the corresponding object information, thereby increasing data diversity without requiring additional manual effort or time investment.
Data Source
AI summary
The present invention discloses an image data augmentation method that includes the steps outlined below. At least one distortion operation function is retrieved. A plurality of pixels included in the image are twisted according to the distortion operation function to generate at least one augmented image. Object information of each of at least one object included in the image is converted according to the distortion operation function to generate object information conversion result. The augmented image, a class tag of each of the at least one object and the object information conversion result are fed to a machine learning module to generate a machine learning result.