Evaluating the Impact of Data Anonymization on Image Retrieval

TL;DR

This paper systematically evaluates how different data anonymization techniques impact the performance of content-based image retrieval systems, providing insights for balancing privacy and retrieval accuracy.

Contribution

It introduces a systematic evaluation framework and assesses multiple anonymization methods and strategies on several datasets, revealing biases and performance impacts.

Findings

Models trained on original data yield more similar retrievals after anonymization.

Anonymization introduces a bias favoring models trained on unaltered data.

The study offers practical insights for developing privacy-preserving CBIR systems.

Abstract

With the growing importance of privacy regulations such as the General Data Protection Regulation, anonymizing visual data is becoming increasingly relevant across institutions. However, anonymization can negatively affect the performance of Computer Vision systems that rely on visual features, such as Content-Based Image Retrieval (CBIR). Despite this, the impact of anonymization on CBIR has not been systematically studied. This work addresses this gap, motivated by the DOKIQ project, an artificial intelligence-based system for document verification actively used by the State Criminal Police Office Baden-W\"urttemberg. We propose a simple evaluation framework: retrieval results after anonymization should match those obtained before anonymization as closely as possible. To this end, we systematically assess the impact of anonymization using two public datasets and the internal DOKIQ dataset. Our experiments span three anonymization methods, four anonymization degrees, and four training strategies, all based on the state of the art backbone Self-Distillation with No Labels (DINO)v2. Our results reveal a pronounced retrieval bias in favor of models trained on original data, which produce the most similar retrievals after anonymization. The findings of this paper offer practical insights for developing privacy-compliant CBIR systems while preserving performance.