Problem
Research questions and friction points this paper is trying to address.
Child Sexual Abuse Imagery
privacy-preserving
dataset sharing
computer vision
ethical constraints
CSA-Graphs: A Privacy-Preserving Structural Dataset for Child Sexual Abuse Research
🤖 AI Summary
Research on child sexual abuse imagery (CSAI) faces significant challenges in data sharing due to stringent legal and ethical constraints, which severely impedes reproducibility and progress. To address this, this work proposes CSA-Graphs, the first privacy-preserving dataset built upon a dual-modality graph structure. It models semantic object relationships using scene graphs and encodes human pose information via skeletal graphs, effectively removing all sensitive visual content while retaining discriminative contextual features. By integrating these two modalities through graph neural networks, experiments demonstrate that both graph representations independently support effective CSAI classification, with their fusion yielding substantially improved performance. This approach establishes a compliant and reproducible foundation for future research in this critical domain.
📝 Abstract
Child Sexual Abuse Imagery (CSAI) classification is an important yet challenging problem for computer vision research due to the strict legal and ethical restrictions that prevent the public sharing of CSAI datasets. This limitation hinders reproducibility and slows progress in developing automated methods. In this work, we introduce CSA-Graphs, a privacy-preserving structural dataset. Instead of releasing the original images, we provide structural representations that remove explicit visual content while preserving contextual information. CSA-Graphs includes two complementary graph-based modalities: scene graphs describing object relationships and skeleton graphs encoding human pose. Experiments show that both representations retain useful information for classifying CSAI, and that combining them further improves performance. This dataset enables broader research on computer vision methods for child safety while respecting legal and ethical constraints.
Innovation
Methods, ideas, or system contributions that make the work stand out.
privacy-preserving
graph-based representation
scene graph
skeleton graph
CSAI classification
🔎 Similar Papers
No similar papers found.
