Update README.md
Browse files
README.md
CHANGED
|
@@ -8,6 +8,13 @@ base_model:
|
|
| 8 |
---
|
| 9 |
Checks whether an image is real or fake (AI-generated).
|
| 10 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
See https://www.kaggle.com/code/dima806/deepfake-vs-real-faces-detection-vit for more details.
|
| 12 |
|
| 13 |
```
|
|
|
|
| 8 |
---
|
| 9 |
Checks whether an image is real or fake (AI-generated).
|
| 10 |
|
| 11 |
+
**Note to users who want to use this model in production**
|
| 12 |
+
Beware that this model is trained on a dataset collected about 3 years ago.
|
| 13 |
+
Since then, there is a remarkable progress in generating deepfake images with common AI tools, resulting in a significant concept drift.
|
| 14 |
+
To mitigate that, I urge you to retrain the model using the latest available labeled data.
|
| 15 |
+
As a quick-fix approach, simple reducing the threshold (say from default 0.5 to 0.1 or even 0.01) of labelling image as a fake may suffice.
|
| 16 |
+
However, you will do that at your own risk, and retraining the model is the better way of handling the concept drift.
|
| 17 |
+
|
| 18 |
See https://www.kaggle.com/code/dima806/deepfake-vs-real-faces-detection-vit for more details.
|
| 19 |
|
| 20 |
```
|