Update README.md
Browse files
README.md
CHANGED
|
@@ -1,7 +1,19 @@
|
|
| 1 |
---
|
| 2 |
license: mit
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
| 4 |
|
| 5 |
|
| 6 |
An embedding model to classify images into FLUX generated images and non-flux photographs.
|
| 7 |
-
The embeddings are 128 dimensional and can be used in another classifier to classify.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: mit
|
| 3 |
+
library_name: keras
|
| 4 |
+
pipeline_tag: image-classification
|
| 5 |
+
tags:
|
| 6 |
+
- image classification
|
| 7 |
+
- embeddings
|
| 8 |
---
|
| 9 |
|
| 10 |
|
| 11 |
An embedding model to classify images into FLUX generated images and non-flux photographs.
|
| 12 |
+
The embeddings are 128 dimensional and can be used in another classifier to classify.
|
| 13 |
+
|
| 14 |
+
The model can load Fourier transformed images of size 512x512 which are then fed into the model and a 128 length output vector is produced.
|
| 15 |
+
The steps to create the embeddings can be described as:
|
| 16 |
+
1. Resize the images to 512x512.
|
| 17 |
+
2. Transform the images into their Fourier image.
|
| 18 |
+
3. Input the images into the model using predict.
|
| 19 |
+
4. The output will be a 128-length vector for use in classification models.
|