CoChat
Sign Up
Cited by 29Privacy-Preserving Technologies in DataAdvanced Neural Network ApplicationsMachine Learning and Data Classification

The DOI matches its Crossref record; 8 of the 8 references it cites were found in a registry or an index.

Balancing Privacy and Accuracy: Exploring the Impact of Data Anonymization on Deep Learning Models in Computer Vision

Open access
Lee, Jun Ha, You, Su Jeong·2024·IEEE Access
DOI 10.1109/access.2024.3352146
Open this paper in CoChat — free accountOpen PDF

Abstract

Computer vision has become indispensable in various applications, including autonomous driving, medical imaging, security and surveillance, robotics, and pattern recognition. In recent years, the quality of training data has emerged as a critical factor for ensuring effectiveness in real-world scenarios. However, the increasing stringency of privacy regulations in various regions necessitates careful handling of collected images for computer vision. Personal information within images is typically anonymized by applying anonymization patterns to remove it. Empirical findings underscore the significant influence of data quality on the training of deep learning models. Striking the right balance between privacy and recognition performance becomes paramount. Therefore, it is essential to understand how the anonymization of image datasets affects deep learning model performance. In this paper, we thoroughly analyze the effects of different anonymization techniques on the performance of deep learning-based models in computer vision tasks, with a particular emphasis on presenting a model-centric perspective, such as the type of deep learning model and the number of parameters. We aim to provide valuable insights and guidelines for selecting the optimal level of anonymization that strikes a balance between recognition accuracy and privacy protection.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1109/access.2024.3352146

Cites

8 of 8 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Flagged first, then by citations

Show all 8 references

Research works printed but not linked: 65

  1. 10.1016/j.cmpb.2021.106504 · DOI · Paper page
  2. 10.3390/app11020796 · DOI · Paper page
  3. 2016, Regulation (EU) 2016/679 of the European Parliament and of the Council of 27, Apr. 2016 on the Protection of Natural Persons With Regard to the Processing of Personal Data and on the Free Movement of Such Data, and Repealing, Directive 95/46/EC (General Data Protection Regulation)
  4. 2020, California Consumer Privacy Act of 2018 (as Amended by the California Privacy Rights Act of 2020)
  5. 10.1109/TPAMI.1983.4767365 · DOI
  6. 10.1109/CVPR42600.2020.00252 · DOI · Paper page
  7. 10.1109/AI4I46381.2019.00016 · DOI
  8. 10.1109/FG52635.2021.9666936 · DOI
  9. Yang, A study of face obfuscation in ImageNet, 25313
  10. 10.1109/CVPRW53098.2021.00364 · DOI
  11. 10.1007/s12652-015-0303-x · DOI
  12. Lubarsky, 2020, Re-Identification of ‘Anonymized’ Data. Georgetown Law Technology Review
  13. 10.1007/978-3-319-10602-1_48 · DOI · Paper page
  14. 10.1109/DSN.2016.40 · DOI
  15. 10.1109/TKDE.2005.32 · DOI
  16. 10.1109/ICDAR.2003.1227801 · DOI
  17. 10.1109/CVPRW59228.2023.00019 · DOI
  18. 10.1109/EDCC.2017.17 · DOI
  19. Dvo ek, What is the cost of privacy?, 1602, 696
  20. 10.1109/IV51971.2022.9827262 · DOI
  21. 10.1109/WACV56688.2023.00148 · DOI
  22. 10.1109/WACV56688.2023.00138 · DOI
  23. 10.1145/3422622 · DOI · Paper page
  24. Li, 2019, FaceShifter: Towards high fidelity and occlusion aware face swapping, arXiv:1912.13457 · arXiv
  25. 10.1109/CVPR46437.2021.00232 · DOI
  26. 10.1109/CVPR42600.2020.00926 · DOI · Paper page
  27. Karras, Progressive growing of GANs for improved quality stability and variation, 1
  28. 10.1109/TCSVT.2021.3066054 · DOI
  29. 10.3390/s21010058 · DOI
  30. 10.1109/ICDAR.2011.100 · DOI
  31. 10.1109/CVPR.2016.350 · DOI · Paper page
  32. 10.1007/0-306-47044-6_11 · DOI
  33. 10.1007/11558484_18 · DOI
  34. 10.1109/AFGR.2002.1004148 · DOI
  35. 10.1109/MIPRO.2014.6859767 · DOI
  36. 10.1109/AFGR.2002.1004181 · DOI
  37. 10.1109/TIP.2020.3018224 · DOI
  38. 10.1109/ICCV48922.2021.01437 · DOI
  39. 10.1117/12.2524023 · DOI
  40. 10.1109/WACV56688.2023.00126 · DOI
  41. 10.1007/978-3-319-46475-6_53 · DOI
  42. Lyu, 2022, RTMDet: An empirical study of designing real-time object detectors, arXiv:2212.07784 · arXiv
  43. O’Shea, 2015, An introduction to convolutional neural networks, arXiv:1511.08458 · arXiv
  44. Dosovitskiy, 2020, An image is worth 16✗16 words: Transformers for image recognition at scale, arXiv:2010.11929 · arXiv
  45. 2023, YOLOv8
  46. 10.1109/TKDE.2009.191 · DOI · Paper page
  47. 10.1016/j.aiopen.2021.08.002 · DOI · Paper page
  48. 10.1109/ACCESS.2020.3034343 · DOI
  49. 10.1109/CVPR.2014.81 · DOI · Paper page
  50. 10.1007/s11263-009-0275-4 · DOI · Paper page
  51. 10.1109/CVPR.2011.5995316 · DOI
  52. 10.1109/CVPR.2016.319 · DOI · Paper page
  53. 10.1109/ICCV.2017.74 · DOI · Paper page
  54. 10.1007/s10115-021-01605-0 · DOI
  55. Zhang, 2018, A study on overfitting in deep reinforcement learning, arXiv:1804.06893 · arXiv
  56. 10.1109/TPAMI.2019.2929257 · DOI · Paper page
  57. 10.1609/aaai.v36i1.19880 · DOI
  58. 10.1145/3524497 · DOI
  59. Sung, Training neural networks with fixed sparse masks, 24193
  60. 10.3390/electronics12081789 · DOI
  61. Brownlee, 2018, How to avoid overfitting in deep learning neural networks, Mach. Learn. Mastery, 17, 12
  62. Bai, Are transformers more robust than CNNs?, 26831
  63. Bochkovskiy, 2020, YOLOv4: Optimal speed and accuracy of object detection, arXiv:2004.10934 · arXiv
  64. 10.1109/ICCV.2017.324 · DOI · Paper page
  65. Balasubramaniam, 2022, Object detection in autonomous vehicles: Status and open challenges, arXiv:2201.07706 · arXiv

Cited by

Sorted by citations

See all 29

Metadata

Authors
Lee, J. H., & You, S. J.
Affiliations
Korea Institute of Industrial Technology (KR)
ORCID
Venue
IEEE Access
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Published
2024
Volume
12
Pages
8346-8358
ISSN
2169-3536 (ISSN-L)
DOI
10.1109/access.2024.3352146
OpenAlex
W4390691711
Cited by
29
References
8
Citations by year
2026: 5 · 2025: 15 · 2024: 9
FWCI
4.30
Type
journal article
Language
en
Open access
Licence
Locations
  • IEEE Access (publishedVersion, cc-by-nc-nd)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • Institute of Information and Communication Planning and Evaluation (IITP) Grant
  • Korean Government [Ministry of Science and ICT (MSIT)], Development of Personal Information Processing Technology for Data Collection/Utilization for Autonomous Driving (2021-0-01062)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
  • Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
Keywords
data anonymization, image anonymization, privacy-preserving machine learning, recognition accuracy, privacy-accuracy trade-off, model size, privacy regulation, training data quality

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review