docs: Update README
Browse files
README.md
CHANGED
@@ -1,3 +1,63 @@
|
|
1 |
-
---
|
2 |
-
license: bsd-3-clause
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: bsd-3-clause
|
3 |
+
base_model:
|
4 |
+
- microsoft/resnet-50
|
5 |
+
pipeline_tag: image-feature-extraction
|
6 |
+
---
|
7 |
+
|
8 |
+
# ResNet-50 Embeddings Only
|
9 |
+
|
10 |
+
This is a modified version of a standard ResNet-50 architecture, where the final, fully connected layer that does the classification, has been removed.
|
11 |
+
|
12 |
+
This effectively gives you the embeddings.
|
13 |
+
|
14 |
+
NB: You may want to flatten the embeddings, as it'll be of shape `(1, 20248, 1, 1)` otherwise.
|
15 |
+
|
16 |
+
# Example
|
17 |
+
|
18 |
+
```python
|
19 |
+
import onnxruntime
|
20 |
+
from PIL import Image
|
21 |
+
from torchvision import transforms
|
22 |
+
|
23 |
+
|
24 |
+
def load_and_preprocess_image(image_path):
|
25 |
+
# Define the same preprocessing as used in training
|
26 |
+
preprocess = transforms.Compose(
|
27 |
+
[
|
28 |
+
transforms.Resize(256),
|
29 |
+
transforms.CenterCrop(224),
|
30 |
+
transforms.ToTensor(),
|
31 |
+
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
|
32 |
+
]
|
33 |
+
)
|
34 |
+
|
35 |
+
# Open the image file
|
36 |
+
img = Image.open(image_path)
|
37 |
+
|
38 |
+
# Preprocess the image
|
39 |
+
img_preprocessed = preprocess(img)
|
40 |
+
|
41 |
+
# Add batch dimension
|
42 |
+
return img_preprocessed.unsqueeze(0).numpy()
|
43 |
+
|
44 |
+
|
45 |
+
onnx_model_path = "resnet50_embeddings.onnx"
|
46 |
+
|
47 |
+
session = onnxruntime.InferenceSession(onnx_model_path)
|
48 |
+
|
49 |
+
input_name = session.get_inputs()[0].name
|
50 |
+
|
51 |
+
# Load and preprocess an image (replace with your image path)
|
52 |
+
image_path = "disco-ball.jpg"
|
53 |
+
input_data = load_and_preprocess_image(image_path)
|
54 |
+
|
55 |
+
# Run inference
|
56 |
+
outputs = session.run(None, {input_name: input_data})
|
57 |
+
|
58 |
+
# The output should be a single tensor (the embeddings)
|
59 |
+
embeddings = outputs[0]
|
60 |
+
|
61 |
+
# Flatten the embeddings
|
62 |
+
embeddings = embeddings.reshape(embeddings.shape[0], -1)
|
63 |
+
```
|