Conditional Adversarial Camera Model Anonymization
This addresses the problem of privacy protection in digital images by anonymizing camera models, though it appears incremental as it builds on prior work by extending transformations to include low-frequency information.
The paper tackled the problem of camera model anonymization by transforming both high and low spatial frequency artifacts to change the apparent capture model, achieving efficacy in a restrictive non-interactive black-box setting as demonstrated by quantitative comparisons.
The model of camera that was used to capture a particular photographic image (model attribution) is typically inferred from high-frequency model-specific artifacts present within the image. Model anonymization is the process of transforming these artifacts such that the apparent capture model is changed. We propose a conditional adversarial approach for learning such transformations. In contrast to previous works, we cast model anonymization as the process of transforming both high and low spatial frequency information. We augment the objective with the loss from a pre-trained dual-stream model attribution classifier, which constrains the generative network to transform the full range of artifacts. Quantitative comparisons demonstrate the efficacy of our framework in a restrictive non-interactive black-box setting.