DeepFreqMark: End-To-End Learnable Frequency-Domain Watermarking with Spherical Attack Simulation for Latent Diffusion Models
2026-08-10 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial Intelligence
AI summaryⓘ
The authors address issues with hidden watermarks in AI-generated images made by Latent Diffusion Models, which help prove ownership but have limits in how much data they can hide. They created DeepFreqMark, a new system that uses neural networks to encode and decode watermarks in a smarter way, replacing older fixed patterns. To train their system efficiently, they invented a method mimicking likely attacks on these watermarks without heavy computation. Their tests show DeepFreqMark is better at recovering hidden data accurately even after attacks and can hide more information than previous methods.
Latent Diffusion ModelsAI-generated imageswatermarkingfrequency-domain watermarkingneural message encoderDDIM inversionSpherical Linear Interpolation (Slerp)Bit Error RateGaussian varianceattack simulation
Authors
Chen-Hsiu Huang, Mario Köppen, Ja-Ling Wu
Abstract
The proliferation of AI-generated images produced by Latent Diffusion Models (LDMs) has raised critical concerns regarding copyright infringement and misinformation. Although existing frequency-domain watermarking methods embed handcrafted geometric patterns into the initial latent noise prior to generation, they suffer from limited capacity and rigid pattern designs. We propose DeepFreqMark, an end-to-end learnable frequency-domain watermarking framework that replaces manual pattern engineering with a neural message encoder and decoder. To circumvent the computational bottleneck caused by Denoising Diffusion Implicit Model (DDIM) inversion during training, we introduce a Spherical Linear Interpolation (Slerp)-based attack simulation. This approach operates directly on the noise latent while strictly preserving the Gaussian variance. Extensive experiments demonstrate that DeepFreqMark achieves significantly lower Bit Error Rates (BER) than baseline methods under real-world attacks and scales to 256 bits message capacity. Our source code is available at https://github.com/chenhsiu48/DeepFreqMark.