AnyPattern: Towards In-context Image Copy Detection

Wenhao Wang, Yifan Sun, Zhentao Tan, Yi Yang

Abstract: This paper explores in-context learning for image copy detection (ICD), i.e., prompting an ICD model to identify replicated images with new tampering patterns without the need for additional training. The prompts (or the contexts) are from a small set of image-replica pairs that reflect the new patterns and are used at inference time. Such in-context ICD has good practical value, because it requires no fine-tuning and thus facilitates fast reaction to the emergence of unseen patterns. To accommodate the “seen \(\rightarrow \) unseen” generalization scenario, we construct the first large-scale pattern dataset named AnyPattern, which has the largest number of tampering patterns (90 for training and 10 for testing) among all the existing datasets. We benchmark AnyPattern with popular ICD methods and reveal that existing methods barely generalize to novel patterns. We further propose a simple in-context ICD method named ImageStacker. ImageStacker learns to select the most representative image-replica pairs and employs them as the pattern prompts in a stacking manner (rather than the popular concatenation manner). Experimental results show (1) training with our large-scale dataset substantially benefits pattern generalization (\(+26.66 \%\) \(\mu AP\)), (2) the proposed ImageStacker facilitates effective in-context ICD (another round of \(+16.75 \%\) \(\mu AP\)), and (3) AnyPattern enables in-context ICD, i.e., without such a large-scale dataset, in-context learning does not emerge even with our ImageStacker. Beyond the ICD task, we also demonstrate how AnyPattern can benefit artists, i.e., the pattern retrieval method trained on AnyPattern generalizes to identify style mimicry by text-to-image models. The project is publicly available at https://anypattern.github.io.

External IDs:doi:10.1007/s11263-025-02633-x