Downloads · 30 days
3
5% of all-time downloads
giaupham/fae-hybrid-unet-steganography
fae-hybrid-unet-steganography is a image-to-image model from giaupham. Use it when you need one image transformed into another. It is set up for pytorch. The card lists the license as other.
This repository releases the PyTorch robust-best checkpoints for the manuscript Frequency-Aware Embedding Hybrid U-Net: Robust Image Steganography Against Geometric and Compression Attacks.
Downloads · 30 days
3
5% of all-time downloads
All-time downloads
63
Public
Repo size
244 MB
Likes
0
Public
Click a slice to open those files.
.pth244 MB · 100%
From the Hugging Face model README
This repository releases the PyTorch robust-best checkpoints for the manuscript Frequency-Aware Embedding Hybrid U-Net: Robust Image Steganography Against Geometric and Compression Attacks.
| File | Training domain | Selection | Epoch | SHA-256 |
|---|---|---|---|---|
fae_hybrid_unet_bossbase_robust_best.pth | BOSSBase | robust-best | 209 | DA7FB64B24F29CD830C8858193D614B769271072A7B5BEE5DD051F9BDF6ACDFC |
fae_hybrid_unet_ms_coco_robust_best.pth | MS-COCO | robust-best | 209 | D13C8F5F345516139EC6FC2954343443537764234ECA898A45EFF8877A658E00 |
Both checkpoints use seed 456, 256 x 256 input images, a 1,024-bit barcode payload, and 8 x 8 barcode modulation with 64-fold spatial redundancy per bit. This equals 0.015625 bits per spatial pixel, or 0.005208 bits per RGB channel when the denominator includes all three image channels.
| Metric | BOSSBase | MS-COCO |
|---|---|---|
| PSNR | 41.03 dB | 42.93 dB |
| SSIM | 0.969 | 0.987 |
| Clean extraction accuracy | 100.00% | 100.00% |
| JPEG Q50 accuracy | 83.99% | 91.44% |
| Rotation 15-degree accuracy | 98.18% | 49.99% |
| Average non-identity attack accuracy | 96.11% | 90.57% |
The BOSSBase and MS-COCO rotation results differ substantially. Claims about geometric robustness should remain dataset qualified.
These checkpoints support research reproducibility and non-production evaluation of robust image steganography. The released demo is intended to make the embedding, attack, residual, and decoding workflow inspectable.
@article{pham2026frequencyaware,
title = {Frequency-Aware Embedding Hybrid U-Net: Robust Image Steganography Against Geometric and Compression Attacks},
author = {Pham, Ngoc-Giau and Tran, Hong-Ngoc and Vo, Phuoc-Hung},
year = {2026},
note = {Manuscript submitted for peer review}
}
The checkpoint files are provided for academic research and evaluation. See LICENSE before redistribution or commercial use.