The advent of storing images on cloud platforms has introduced serious privacy concerns. The images are routinely scanned by machine learning algorithms to determine the contents. Usually the scanning is for marketing purposes but more malevolent purposes include criminal activity and government surveillance. The images are automatically analysed by machine learning algorithms. Notably, deep convolutional neural networks perform very well at identifying image classes. Obviously, the images could be encrypted before storing to cloud platforms and then decrypted after downloading. This would certainly obfuscate the images. However, many users prefer to be able to peruse the images on the cloud platform. This creates a difficult problem in which users prefer images stored in a way so that a human can understand them but machine learning algorithms cannot. This paper proposes a novel technique, termed seam doppelganger, for formatting images using seam carving to identify seams for replacement. The approach degrades typical image classification performance in order to provide privacy while leaving the image human-understandable. Furthermore, the technique can be largely reversed providing a reasonable facsimile of the original image. Using the ImageNet database for birds, we show how the approach degrades a state-of-the-art residual network (ResNet50) for various amounts of seam replacements.
|Title of host publication||ICPRAM 2021 - Proceedings of the 10th International Conference on Pattern Recognition Applications and Methods|
|Editors||Maria De Marsico, Gabriella Sanniti di Baja, Ana Fred|
|Publisher||SCITEPRESS Digital Library|
|Number of pages||7|
|Publication status||Published - 2021|
|Event||10th International Conference on Pattern Recognition Applications and Methods, ICPRAM 2021 - Virtual, Online|
Duration: 4 Feb 2021 → 6 Feb 2021
|Name||ICPRAM 2021 - Proceedings of the 10th International Conference on Pattern Recognition Applications and Methods|
|Conference||10th International Conference on Pattern Recognition Applications and Methods, ICPRAM 2021|
|Period||4/02/21 → 6/02/21|
Bibliographical noteFunding Information:
This work was supported in part by the University of Montevallo Contract #19-0501-001. The authors greatly appreciate the support of the staff involved in the project. Without their efforts this research could not have been conducted.
© 2021 by SCITEPRESS - Science and Technology Publications, Lda. All rights reserved
Copyright 2021 Elsevier B.V., All rights reserved.
- Adversarial perturbations
- Image classification
- Privacy protection
- Seam carving