A Privacy-Preserving Semantic-Segmentation Method Using Domain-Adaptation Technique
This addresses privacy concerns in image analysis for applications like medical imaging or surveillance, but it is incremental as it builds on existing domain-adaptation and ViT techniques.
The paper tackles the problem of privacy-preserving semantic segmentation by applying perceptual encryption to both training and test images, achieving nearly the same accuracy as non-encrypted models, with experimental confirmation using a ViT-based segmentation model.
We propose a privacy-preserving semantic-segmentation method for applying perceptual encryption to images used for model training in addition to test images. This method also provides almost the same accuracy as models without any encryption. The above performance is achieved using a domain-adaptation technique on the embedding structure of the Vision Transformer (ViT). The effectiveness of the proposed method was experimentally confirmed in terms of the accuracy of semantic segmentation when using a powerful semantic-segmentation model with ViT called Segmentation Transformer.