Beyond Blanket Masking: Examining Granularity for Privacy Protection in Images Captured by Blind and Low Vision Users

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Murrugarra-LLerena, Jeffri, Niu, Haoran, Barber, K. Suzanne, Daumé III, Hal, Cao, Yang Trista, Cascante-Bonilla, Paola
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908487547617280
author Murrugarra-LLerena, Jeffri
Niu, Haoran
Barber, K. Suzanne
Daumé III, Hal
Cao, Yang Trista
Cascante-Bonilla, Paola
author_facet Murrugarra-LLerena, Jeffri
Niu, Haoran
Barber, K. Suzanne
Daumé III, Hal
Cao, Yang Trista
Cascante-Bonilla, Paola
contents As visual assistant systems powered by visual language models (VLMs) become more prevalent, concerns over user privacy have grown, particularly for blind and low vision users who may unknowingly capture personal private information in their images. Existing privacy protection methods rely on coarse-grained segmentation, which uniformly masks entire private objects, often at the cost of usability. In this work, we propose FiGPriv, a fine-grained privacy protection framework that selectively masks only high-risk private information while preserving low-risk information. Our approach integrates fine-grained segmentation with a data-driven risk scoring mechanism. We evaluate our framework using the BIV-Priv-Seg dataset and show that FiG-Priv preserves +26% of image content, enhancing the ability of VLMs to provide useful responses by 11% and identify the image content by 45%, while ensuring privacy protection. Project Page: https://artcs1.github.io/VLMPrivacy/
format Preprint
id arxiv_https___arxiv_org_abs_2508_09245
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Beyond Blanket Masking: Examining Granularity for Privacy Protection in Images Captured by Blind and Low Vision Users
Murrugarra-LLerena, Jeffri
Niu, Haoran
Barber, K. Suzanne
Daumé III, Hal
Cao, Yang Trista
Cascante-Bonilla, Paola
Computer Vision and Pattern Recognition
As visual assistant systems powered by visual language models (VLMs) become more prevalent, concerns over user privacy have grown, particularly for blind and low vision users who may unknowingly capture personal private information in their images. Existing privacy protection methods rely on coarse-grained segmentation, which uniformly masks entire private objects, often at the cost of usability. In this work, we propose FiGPriv, a fine-grained privacy protection framework that selectively masks only high-risk private information while preserving low-risk information. Our approach integrates fine-grained segmentation with a data-driven risk scoring mechanism. We evaluate our framework using the BIV-Priv-Seg dataset and show that FiG-Priv preserves +26% of image content, enhancing the ability of VLMs to provide useful responses by 11% and identify the image content by 45%, while ensuring privacy protection. Project Page: https://artcs1.github.io/VLMPrivacy/
title Beyond Blanket Masking: Examining Granularity for Privacy Protection in Images Captured by Blind and Low Vision Users
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.09245