DNA: Dual-stage Native Attribution for Generated Image Source Tracing

2026-07-15

AI Summary

This work addresses the challenge of effectively tracing generated images to their source when confronted with unknown origins and highly similar variants of generative models. The authors propose a novel two-stage, training-free native attribution framework: it first performs family-level coarse filtering via Autoencoder Double Reconstruction (AEDR), followed by variant-level fine-grained attribution through Native Prediction Consistency (NPC). This approach is the first to reveal the inherent hierarchical nature of attribution signals across architectural levels within generative models, leveraging this insight to establish an unsupervised, open-set attribution mechanism with strong generalization capabilities. Evaluated on the DNA-30K benchmark, the method achieves an end-to-end accuracy of 89.11%, substantially outperforming the strongest baseline by 55.30% and vastly exceeding random guessing (<1%).

Abstract

The rapid evolution of image generation has produced numerous within-family variants, making source-model attribution of suspect images increasingly important for digital forensics. Existing proactive methods rely on watermark embedding or model modification, which may degrade visual quality and limit deployment flexibility. Passive methods often rely on large-scale supervised training or a single reconstruction signal, limiting their ability to handle unknown sources and distinguish highly similar within-family variants. We observe that attribution signals in latent generative models are naturally stratified across architectural levels: VAE-level cues reflect family-shared information, whereas backbone-level cues capture variant-specific behaviors. Motivated by this insight, we propose Dual-stage Native Attribution (DNA), a coarse-to-fine framework that follows this hierarchy without additional neural-network training. The coarse-grained stage uses Autoencoder Double-Reconstruction (AEDR) for efficient open-set family-level screening. The fine-grained stage performs closed-set model-level attribution with Native Prediction Consistency (NPC), which compares native prediction errors of within-family variants across multiple noise levels under semantic conditioning and attributes the source via normalized calibrated scores. To enable systematic evaluation, we construct DNA-30K, a benchmark for within-family variant attribution under open-set family-level evaluation. It comprises 30,000 images generated by 24 candidate models across six families spanning both denoising diffusion and flow matching, plus non-candidate generated and natural images as unknown sources. Experiments show that DNA achieves 89.11% end-to-end attribution accuracy on a task where random guessing accuracy is below 1% and outperforms the strongest baseline by 33.81% even when AEDR is used as the coarse-grained stage.

Problem

• source attribution
• generated image
• within-family variants
• digital forensics
• open-set evaluation

Innovation

• Dual-stage Native Attribution
• Autoencoder Double-Reconstruction
• Native Prediction Consistency
• within-family variant attribution
• open-set source tracing