Image anonymization is widely adapted in practice to comply with privacy regulations in many regions. However, anonymization often degrades the quality of the data, reducing its utility for computer vision development. In this paper, we investigate the impact of image anonymization for training computer vision models on key computer vision tasks (detection, instance segmentation, and pose estimation). Specifically, we benchmark the recognition drop on common detection datasets, where we evaluate both traditional and realistic anonymization for faces and full bodies. Our comprehensive experiments reflect that traditional image anonymization substantially impacts final model performance, particularly when anonymizing the full body. Furthermore, we find that realistic anonymization can mitigate this decrease in performance, where our experiments reflect a minimal performance drop for face anonymization. Our study demonstrates that realistic anonymization can enable privacy-preserving computer vision development with minimal performance degradation across a range of important computer vision benchmarks.

翻译：图像匿名化在许多地区被广泛采用以符合隐私法规要求。然而，匿名化往往会降低数据质量，从而削弱其在计算机视觉开发中的实用性。本文研究了图像匿名化对关键计算机视觉任务（目标检测、实例分割和姿态估计）中模型训练的影响。具体而言，我们在常见检测数据集上基准测试了识别性能下降情况，评估了针对人脸和全身的传统及现实匿名化方法。综合实验表明，传统图像匿名化会显著影响最终模型性能，尤其在全身匿名化时更为突出。进一步研究发现，现实匿名化可缓解这种性能下降，其中人脸匿名化仅导致极小的性能损失。本研究证明，现实匿名化能够在保持隐私保护的同时，在多个重要计算机视觉基准测试中以最小性能损失实现模型开发。