VPD-100K: Towards Generalizable and Fine-grained Visual Privacy Protection

Privacy protection has become a critical requirement in the era of ubiquitous visual data sharing, imposing higher demands on efficient and robust privacy detection algorithms. However, current robust detection models are severely hindered by the lack of comprehensive datasets. Existing privacy-oriented datasets often suffer from limited scale, coarse-grained annotations, and narrow domain coverage, failing to capture the intricate details of sensitive information in real-world environments. To bridge this gap, we present a large-scale, fine-grained Visual Privacy Dataset (VPD-100K), designed to facilitate generalized privacy detection. We establish a holistic taxonomy comprising four primary domains: Human Presence, On-Screen Personally Identifiable Information (PII), Physical Identifiers, and Location Indicators, containing 100,000 images annotated with 33 fine-grained classes and over 190,000 object instances. Statistical analysis reveals that our dataset features long-tailed distributions, small object scales, and high visual complexity. These characteristics make the dataset particularly valuable for demanding, unconstrained applications such as live streaming, where actors frequently face unintentional, real-time information leakage. Furthermore, we design an effective frequency-enhance lightweight module consisting of frequency-domain attention fusion and adaptive spectral gating mechanism that breaks the limitations of spatial pixel intensity to better capture the subtle details of sensitive information. Extensive experiments conducted on both diverse image and streaming videos benchmarks consistently demonstrate the effectiveness of our VPD-100K dataset and the well-curated frequency mechanism.

(1) Problem: Sharing photos and live streams online often exposes private information accidentally, but current artificial intelligence tools struggle to spot small or hidden sensitive details in complex backgrounds. Existing datasets used to train these tools are typically too small or lack the specific examples needed to handle real-world scenarios, such as tiny on-screen text or distant ID cards. (2) Solution: To tackle this, we introduce the Visual Privacy Dataset (VPD-100K), a massive collection of 100,000 images covering detailed privacy areas like human presence, location indicators, physical identifiers, and on-screen personal information. Alongside this dataset, we design a specialized, lightweight AI tool that analyzes high-frequency image textures—rather than just basic spatial pixels—to successfully detect even the most subtle or "camouflaged" sensitive details. (3) Impact: This research provides a powerful foundation for creating highly accurate, real-time privacy shields. Ultimately, it helps platforms and content creators safely navigate the digital world by automatically preventing unintentional information leaks during live, unconstrained broadcasting.