Privacy-preserving Object Detection
This addresses privacy and bias issues for the computer vision community, but it is incremental as it applies existing methods to a known dataset.
The paper tackled privacy and bias in object detection by showing that anonymizing COCO datasets through face blurring and balanced face swapping retains detection performance while preserving privacy and partially balancing bias, with specific performance metrics implied but not detailed.
Privacy considerations and bias in datasets are quickly becoming high-priority issues that the computer vision community needs to face. So far, little attention has been given to practical solutions that do not involve collection of new datasets. In this work, we show that for object detection on COCO, both anonymizing the dataset by blurring faces, as well as swapping faces in a balanced manner along the gender and skin tone dimension, can retain object detection performances while preserving privacy and partially balancing bias.