Tracing Copied Pixels and Regularizing Patch Affinity in Copy Detection

2026-02-19

AI Summary

Existing self-supervised image copy detection methods struggle to establish fine-grained correspondences under complex edits, limiting their performance. This work proposes PixTrace, a module that explicitly tracks the geometric traceability of pixel coordinates throughout editing transformations, jointly learning pixel-level spatial mappings and patch-level similarity. To enhance the reliability of self-supervised signals, we introduce CopyNCE loss, which regularizes patch affinities using overlap ratios. Evaluated on DISC21, our method achieves state-of-the-art performance, with 88.7% uAP / 83.9% RP90 for the matcher and 72.6% uAP / 68.4% RP90 for the descriptor, demonstrating superior robustness, accuracy, and interpretability.

Technology Category

• Computer Vision: Motion & Tracking
• Search and Optimization: Learning to Search
• Machine Learning: Unsupervised & Self-Supervised Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Topic discovery and tracking

Abstract

Image Copy Detection (ICD) aims to identify manipulated content between image pairs through robust feature representation learning. While self-supervised learning (SSL) has advanced ICD systems, existing view-level contrastive methods struggle with sophisticated edits due to insufficient fine-grained correspondence learning. We address this limitation by exploiting the inherent geometric traceability in edited content through two key innovations. First, we propose PixTrace - a pixel coordinate tracking module that maintains explicit spatial mappings across editing transformations. Second, we introduce CopyNCE, a geometrically-guided contrastive loss that regularizes patch affinity using overlap ratios derived from PixTrace's verified mappings. Our method bridges pixel-level traceability with patch-level similarity learning, suppressing supervision noise in SSL training. Extensive experiments demonstrate not only state-of-the-art performance (88.7% uAP / 83.9% RP90 for matcher, 72.6% uAP / 68.4% RP90 for descriptor on DISC21 dataset) but also better interpretability over existing methods.

Problem

• Image Copy Detection
• self-supervised learning
• fine-grained correspondence
• manipulated content
• contrastive learning

Innovation

• PixTrace
• CopyNCE
• pixel coordinate tracking
• geometrically-guided contrastive learning
• copy detection