chore: populate readme
Browse files
README.md
CHANGED
|
@@ -2,10 +2,62 @@
|
|
| 2 |
tags:
|
| 3 |
- model_hub_mixin
|
| 4 |
- pytorch_model_hub_mixin
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
license: cc-by-nc-4.0
|
| 6 |
---
|
| 7 |
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
tags:
|
| 3 |
- model_hub_mixin
|
| 4 |
- pytorch_model_hub_mixin
|
| 5 |
+
- watermarking
|
| 6 |
+
- latent-diffusion
|
| 7 |
+
- stable-signature
|
| 8 |
+
- watermark-extraction
|
| 9 |
+
- computer-vision
|
| 10 |
+
- research
|
| 11 |
+
- non-commercial
|
| 12 |
+
- pytorch
|
| 13 |
license: cc-by-nc-4.0
|
| 14 |
---
|
| 15 |
|
| 16 |
+
# **MsgExtractor — Stable Signature Decoder**
|
| 17 |
+
|
| 18 |
+
**A version of the Stable Signature decoder from the Meta AI project “Stable Signature: Rooting Watermarks in Latent Diffusion Models.”**
|
| 19 |
+
|
| 20 |
+
---
|
| 21 |
+
|
| 22 |
+
## Model Summary
|
| 23 |
+
|
| 24 |
+
* **Model Type:** Custom PyTorch Model
|
| 25 |
+
* **Task:** Watermark extraction from watermarked images
|
| 26 |
+
* **Source:** Derived from [facebookresearch/stable_signature](https://github.com/facebookresearch/stable_signature)
|
| 27 |
+
* **License:** **CC-BY-NC 4.0** (Attribution + Non-Commercial)
|
| 28 |
+
* **Framework:** PyTorch
|
| 29 |
+
* **Weights:** Ported from the original TorchScript decoder
|
| 30 |
+
* **Architecture:**
|
| 31 |
+
|
| 32 |
+
* `HiddenDecoder(num_blocks, num_bits, channels, redundancy)`
|
| 33 |
+
* `MsgExtractor(hidden_decoder, in_features, out_features)`
|
| 34 |
+
* **Status:** Research-only, non-commercial
|
| 35 |
+
|
| 36 |
+
### **Load the model**
|
| 37 |
+
|
| 38 |
+
```python
|
| 39 |
+
from modeling_msg_extractor import MsgExtractor
|
| 40 |
+
import torch
|
| 41 |
+
|
| 42 |
+
model = MsgExtractor.from_pretrained("ESmike/StableSignatureDecoder")
|
| 43 |
+
model.eval()
|
| 44 |
+
|
| 45 |
+
img = torch.randn(1, 3, 256, 256) # example input
|
| 46 |
+
bits = model(img)
|
| 47 |
+
print(bits.shape)
|
| 48 |
+
```
|
| 49 |
+
|
| 50 |
+
### Citation
|
| 51 |
+
|
| 52 |
+
If you use this model, you must cite the original Stable Signature paper:
|
| 53 |
+
|
| 54 |
+
```bibtex
|
| 55 |
+
@inproceedings{Fernandez2023StableSignature,
|
| 56 |
+
title={The Stable Signature Rooting Watermarks in Latent Diffusion Models},
|
| 57 |
+
author={Fernandez, Pierre and Chappelier, Vivien and Nguyen-Hong, Son},
|
| 58 |
+
year={2023},
|
| 59 |
+
institution={Meta AI},
|
| 60 |
+
note={Original implementation at https://github.com/facebookresearch/stable_signature}
|
| 61 |
+
}
|
| 62 |
+
```
|
| 63 |
+
|