Region of Interest Loss for Anonymizing Learned Image Compression

TL;DR

This paper introduces a region of interest loss function for learned image compression that anonymizes human faces while maintaining detection of persons, enabling privacy-preserving data transmission in surveillance applications.

Contribution

It proposes a novel ROI-based loss for autoencoders that achieves face anonymization during compression without losing person detectability, integrated into the compression process.

Findings

Faces become unrecognizable after compression with ROI loss.

Persons remain detectable with high accuracy using pre-trained models.

The method balances anonymization quality with compression rate and latency.

Abstract

The use of AI in public spaces continually raises concerns about privacy and the protection of sensitive data. An example is the deployment of detection and recognition methods on humans, where images are provided by surveillance cameras. This results in the acquisition of great amounts of sensitive data, since the capture and transmission of images taken by such cameras happens unaltered, for them to be received by a server on the network. However, many applications do not explicitly require the identity of a given person in a scene; An anonymized representation containing information of the person's position while preserving the context of them in the scene suffices. We show how using a customized loss function on region of interests (ROI) can achieve sufficient anonymization such that human faces become unrecognizable while persons are kept detectable, by training an end-to-end optimized autoencoder for learned image compression that utilizes the flexibility of the learned analysis and reconstruction transforms for the task of mutating parts of the compression result. This approach enables compression and anonymization in one step on the capture device, instead of transmitting sensitive, nonanonymized data over the network. Additionally, we evaluate how this anonymization impacts the average precision of pre-trained foundation models on detecting faces (MTCNN) and humans (YOLOv8) in comparison to non-ANN based methods, while considering compression rate and latency.