Privacy-Preserving Semantic Segmentation without Key Management

TL;DR

This paper introduces a new privacy-preserving semantic segmentation approach that encrypts images with independent keys for each client, enabling secure training and inference without key management.

Contribution

It presents a novel encryption-based method allowing secure semantic segmentation with independent keys, maintaining performance on vision transformer models.

Findings

Effective on Cityscapes dataset with vision transformer model SETR.

Encryption method mitigates performance loss during training.

Supports secure inference without key management.

Abstract

This paper proposes a novel privacy-preserving semantic segmentation method that can use independent keys for each client and image. In the proposed method, the model creator and each client encrypt images using locally generated keys, and model training and inference are conducted on the encrypted images. To mitigate performance degradation, an image encryption method is applied to model training in addition to the generation of test images. In experiments, the effectiveness of the proposed method is confirmed on the Cityscapes dataset under the use of a vision transformer-based model, called SETR.