Time-Series Face Anonymization for Consistent Learning 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 the privacy of individuals while maintaining effective feature information, leading to ineffective data generation.
Innovation Solution
An image processing device and method that performs anonymization on input images by replacing faces with those of another person and ensures continuity and consistency of face features, using an image determination unit to verify that the same face is maintained across images, thereby generating effective learning data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If face images are synthesized from multiple persons to protect privacy, then privacy protection is improved, but feature information of the original image is lost
Solution Approach 1:
The patent applies face swapping technology to replace the face in the target image with a face from another person. This creates a copy of the face image that preserves the original image's features (pose, expression, lighting, background) while changing the identity information, thus protecting privacy without losing feature information needed for machine learning training
Solution Approach 2:
The patent performs localized face swapping only on the face region of the image while preserving the rest of the image features. The face swapping is applied selectively to the face area, maintaining the original image's pose, expression, lighting conditions, and background, thus preserving feature information while protecting privacy
2Object-affected harmful factors
If face swapping is performed to protect privacy, then privacy protection is improved, but consistency of face identity across multiple images is lost
Solution Approach 1:
The patent performs face tracking on the original images before face swapping to identify and register the same person across multiple images. This preliminary action ensures that when face swapping is performed, the same swapped face is applied to all images of the same person, maintaining face identity consistency while protecting privacy
Solution Approach 2:
The patent uses face tracking as a feedback mechanism to monitor and identify the same person across multiple images. This feedback information guides the face swapping process to apply consistent face replacements, ensuring that the swapped faces maintain identity consistency across the image sequence while protecting privacy
Data Source
AI summary
Provided is an image processing device including: an image conversion unit that performs an anonymization process on a plurality of input images captured in a time series; and an image determination unit that determines whether the plurality of input images on which the anonymization process has been performed satisfy a predetermined requirement, wherein the image determination unit performs a predetermined process on the plurality of input images on which the anonymization process has been performed in a case where it is determined that the plurality of input images on which the anonymization process has been performed satisfy the predetermined requirement, the anonymization process includes a process of changing a face of a person depicted in the plurality of input images to a face of another person, and the predetermined requirement includes that faces of persons tracked as the same person in the plurality of input images are the same face in each of the plurality of input images on which the anonymization process has been performed.