Evaluating the Impact of Face Anonymization Methods on Computer Vision Tasks: A Trade-Off Between Privacy and Utility
Data anonymization is an essential prerequisite that enables data sharing in a privacy-preserving manner. However, anonymization affects the quality of the data and thus might affect the performance of later conducted data analysis. In this work, we investigate the influence of different image-based...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10804775/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832590321902419968 |
|---|---|
| author | Roland Stenger Steffen Busse Jonas Sander Thomas Eisenbarth Sebastian Fudickar |
| author_facet | Roland Stenger Steffen Busse Jonas Sander Thomas Eisenbarth Sebastian Fudickar |
| author_sort | Roland Stenger |
| collection | DOAJ |
| description | Data anonymization is an essential prerequisite that enables data sharing in a privacy-preserving manner. However, anonymization affects the quality of the data and thus might affect the performance of later conducted data analysis. In this work, we investigate the influence of different image-based subject anonymization methods on the performance of the three common computer vision (CV) tasks: keypoint detection, instance segmentation, and face detection. We compare the anonymization methods on how much they affect the performance of the CV tasks, as well as their degree of anonymization. Furthermore, a re-identification attack scenario is applied. The study findings indicate a clear trade-off between increased privacy and decreased CV performance. Latent diffusion (called L-Diff in the following) shows the best results with minimal performance decrease on all CV tasks with a low recognition rate. However, the retrieval rate of our attack scenario, which measures the re-identification rate, remains high for all anonymization methods, which contrasts with the respective low level of recognition. |
| format | Article |
| id | doaj-art-46f9965135aa4ea89b52ac3896c9a659 |
| institution | Kabale University |
| issn | 2169-3536 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-46f9965135aa4ea89b52ac3896c9a6592025-01-24T00:01:42ZengIEEEIEEE Access2169-35362025-01-0113110701107910.1109/ACCESS.2024.351944110804775Evaluating the Impact of Face Anonymization Methods on Computer Vision Tasks: A Trade-Off Between Privacy and UtilityRoland Stenger0https://orcid.org/0000-0002-7590-7286Steffen Busse1https://orcid.org/0000-0001-7099-6948Jonas Sander2Thomas Eisenbarth3https://orcid.org/0000-0003-1116-6973Sebastian Fudickar4https://orcid.org/0000-0002-3553-5131Institute for Medical Informatics, University of Luebeck, Lübeck, GermanyDepartment for Health Services Research, Division of Assistance Systems and Medical Device Technology, Carl von Ossietzky Universität Oldenburg, Oldenburg, GermanyInstitute for IT Security, University of Luebeck, Lübeck, GermanyInstitute for IT Security, University of Luebeck, Lübeck, GermanyInstitute for Medical Informatics, University of Luebeck, Lübeck, GermanyData anonymization is an essential prerequisite that enables data sharing in a privacy-preserving manner. However, anonymization affects the quality of the data and thus might affect the performance of later conducted data analysis. In this work, we investigate the influence of different image-based subject anonymization methods on the performance of the three common computer vision (CV) tasks: keypoint detection, instance segmentation, and face detection. We compare the anonymization methods on how much they affect the performance of the CV tasks, as well as their degree of anonymization. Furthermore, a re-identification attack scenario is applied. The study findings indicate a clear trade-off between increased privacy and decreased CV performance. Latent diffusion (called L-Diff in the following) shows the best results with minimal performance decrease on all CV tasks with a low recognition rate. However, the retrieval rate of our attack scenario, which measures the re-identification rate, remains high for all anonymization methods, which contrasts with the respective low level of recognition.https://ieeexplore.ieee.org/document/10804775/Image analysisdata privacysynthetic datacomputer visionperformance evaluationface recognition |
| spellingShingle | Roland Stenger Steffen Busse Jonas Sander Thomas Eisenbarth Sebastian Fudickar Evaluating the Impact of Face Anonymization Methods on Computer Vision Tasks: A Trade-Off Between Privacy and Utility IEEE Access Image analysis data privacy synthetic data computer vision performance evaluation face recognition |
| title | Evaluating the Impact of Face Anonymization Methods on Computer Vision Tasks: A Trade-Off Between Privacy and Utility |
| title_full | Evaluating the Impact of Face Anonymization Methods on Computer Vision Tasks: A Trade-Off Between Privacy and Utility |
| title_fullStr | Evaluating the Impact of Face Anonymization Methods on Computer Vision Tasks: A Trade-Off Between Privacy and Utility |
| title_full_unstemmed | Evaluating the Impact of Face Anonymization Methods on Computer Vision Tasks: A Trade-Off Between Privacy and Utility |
| title_short | Evaluating the Impact of Face Anonymization Methods on Computer Vision Tasks: A Trade-Off Between Privacy and Utility |
| title_sort | evaluating the impact of face anonymization methods on computer vision tasks a trade off between privacy and utility |
| topic | Image analysis data privacy synthetic data computer vision performance evaluation face recognition |
| url | https://ieeexplore.ieee.org/document/10804775/ |
| work_keys_str_mv | AT rolandstenger evaluatingtheimpactoffaceanonymizationmethodsoncomputervisiontasksatradeoffbetweenprivacyandutility AT steffenbusse evaluatingtheimpactoffaceanonymizationmethodsoncomputervisiontasksatradeoffbetweenprivacyandutility AT jonassander evaluatingtheimpactoffaceanonymizationmethodsoncomputervisiontasksatradeoffbetweenprivacyandutility AT thomaseisenbarth evaluatingtheimpactoffaceanonymizationmethodsoncomputervisiontasksatradeoffbetweenprivacyandutility AT sebastianfudickar evaluatingtheimpactoffaceanonymizationmethodsoncomputervisiontasksatradeoffbetweenprivacyandutility |