Update README.md
Browse files
README.md
CHANGED
@@ -42,6 +42,24 @@ Detect AI-generated images in:
|
|
42 |
- **Data bias:** Trained primarily on LAION and COCO derivatives ([source][2411.04125v1.pdf])
|
43 |
- **ADDED BY UPLOADER**: Model is already out of date, fails to detect images on newer generation models.
|
44 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
45 |
|
46 |
## How to Use
|
47 |
|
|
|
42 |
- **Data bias:** Trained primarily on LAION and COCO derivatives ([source][2411.04125v1.pdf])
|
43 |
- **ADDED BY UPLOADER**: Model is already out of date, fails to detect images on newer generation models.
|
44 |
|
45 |
+
## Compatibility Notice
|
46 |
+
This repository contains a **Hugging Face transformers-compatible convert** for the original detection methodology from:
|
47 |
+
|
48 |
+
**Original Work**
|
49 |
+
"Community Forensics: Using Thousands of Generators to Train Fake Image Detectors"
|
50 |
+
[arXiv:2411.04125](https://arxiv.org/abs/2411.04125v1) {{Citation from <source_id>2411.04125v1.pdf}}
|
51 |
+
|
52 |
+
**Our Contributions** (Coming soon)
|
53 |
+
⎯ Conversion of original weights to HF format
|
54 |
+
⎯ Added PyTorch inference pipeline
|
55 |
+
⎯ Standardized model card documentation
|
56 |
+
|
57 |
+
**No Training Performed**
|
58 |
+
⎯ Initial model weights sourced from paper authors
|
59 |
+
⎯ No architectural changes or fine-tuning applied
|
60 |
+
|
61 |
+
**Verify Original Performance**
|
62 |
+
Please refer to Table 3 in <source_id data="2411.04125v1.pdf" /> for baseline metrics.
|
63 |
|
64 |
## How to Use
|
65 |
|