PriVisE-Bench: Benchmarking Human-Grounded Contextual Privacy Reasoning of Vision Language Models over Whole-Scene Images
收藏官方服务：
资源简介：
This anonymized dataset accompanies the paper “PriVisE-Bench: Benchmarking Human-Grounded Contextual Privacy Reasoning of Vision Language Models over Whole-Scene Images” for double-blind peer review. It contains the image dataset and metadata as described in the paper. This dataset includes: Image folders: the dataset images, organized by their assigned labels.Metadata_candidate_set.csv: metadata for the complete candidate image set.Metadata_reviewer_set.csv: metadata for the subset selected for independent reviewer validation.Metadata_PriVisE.csv: metadata for the final dataset. Author-identifying information has been omitted to preserve anonymity during the double-blind review process.
提供机构：
Zenodo创建时间：
2026-08-21
