Face Anonymization With Direction Matching for Driving Image Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for generating learning data for machine learning models in automated driving systems fail to protect privacy while maintaining effective feature information, leading to ineffective data generation.
Innovation Solution
An image processing device and method that anonymizes faces by replacing them with those of another person while ensuring the direction information remains consistent, using an image conversion unit and determination unit to ensure the anonymization process meets specific requirements, including matching gaze and face directions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If face images are anonymized by replacing faces with those of another person, then privacy protection is improved, but direction information accuracy deteriorates
Solution Approach 1:
The patent applies local quality by selectively anonymizing only the face region while preserving the direction information (gaze direction and face direction) of the original person. The anonymized face maintains the same directional characteristics as the original face, ensuring that the direction information remains accurate for training purposes while privacy is protected.
Solution Approach 2:
The patent uses copying by creating a synthesized face image that copies the directional characteristics (gaze direction and face direction) from the original face image. The synthesized face is generated to match the original person's direction information while replacing the actual facial features, thus maintaining accuracy for training while protecting privacy.
2Object-affected harmful factors
If face images are converted to protect privacy, then privacy protection is improved, but feature information completeness deteriorates
Solution Approach 1:
The patent applies local quality by performing anonymization only on the face region while preserving other important features such as direction information (gaze direction and face direction). This selective approach ensures that privacy is protected through face anonymization while essential feature information remains complete and available for training.
Solution Approach 2:
The patent uses copying by generating a synthesized face image that copies the directional features from the original image. The synthesized face maintains the same gaze direction and face direction as the original, ensuring that feature information completeness is preserved while privacy is protected through face replacement.
Data Source
AI summary
Provided is an image processing device including: an image conversion unit that performs an anonymization process on an input image; and an image determination unit that determines whether the input image on which the anonymization process has been performed satisfies a predetermined requirement, wherein the image determination unit performs a predetermined process on the input image on which the anonymization process has been performed in a case where it is determined that the input image on which the anonymization process has been performed satisfies the predetermined requirement, the anonymization process includes a process of changing a face of a person depicted in the input image to a face of another person, and the predetermined requirement is that direction information of the face of the person in the input image matches direction information of the face of the other person in the input image on which the anonymization process has been performed.