AlexKoff88
commited on
Commit
·
09ca532
1
Parent(s):
8cea71d
Update README.md
Browse files
README.md
CHANGED
@@ -7,4 +7,40 @@ The model checkpoint trained using https://github.com/AlexKoff88/mobilenetv2_foo
|
|
7 |
The main intend is to use it in samples and demos for model optimization. Here is the advantages:
|
8 |
- FOOD101 can automatically downloaded without registration and SMS.
|
9 |
- It is quite representative to reflect the real world scenarios.
|
10 |
-
- MobileNet v2 is easy to train and lightweight model which is also representative and used in many public benchmarks.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
The main intend is to use it in samples and demos for model optimization. Here is the advantages:
|
8 |
- FOOD101 can automatically downloaded without registration and SMS.
|
9 |
- It is quite representative to reflect the real world scenarios.
|
10 |
+
- MobileNet v2 is easy to train and lightweight model which is also representative and used in many public benchmarks.
|
11 |
+
|
12 |
+
Here is the code to load the checkpoint in PyTorch:
|
13 |
+
|
14 |
+
```python
|
15 |
+
import sys
|
16 |
+
import os
|
17 |
+
|
18 |
+
import torch
|
19 |
+
import torch.nn as nn
|
20 |
+
import torchvision.models as models
|
21 |
+
|
22 |
+
FOOD101_CLASSES = 101
|
23 |
+
|
24 |
+
def fix_names(state_dict):
|
25 |
+
state_dict = {key.replace('module.', ''): value for (key, value) in state_dict.items()}
|
26 |
+
return state_dict
|
27 |
+
|
28 |
+
model = models.mobilenet_v2()
|
29 |
+
|
30 |
+
num_ftrs = model.classifier[1].in_features
|
31 |
+
model.classifier[1] = nn.Linear(num_ftrs, FOOD101_CLASSES)
|
32 |
+
|
33 |
+
if len(sys.argv) > 1:
|
34 |
+
checkpoint_path = sys.argv[1]
|
35 |
+
|
36 |
+
if os.path.isfile(checkpoint_path):
|
37 |
+
print("=> loading checkpoint '{}'".format(checkpoint_path))
|
38 |
+
|
39 |
+
|
40 |
+
checkpoint = torch.load(checkpoint_path)
|
41 |
+
weights = fix_names(checkpoint['state_dict'])
|
42 |
+
model.load_state_dict(weights)
|
43 |
+
|
44 |
+
print("=> loaded checkpoint '{}' (epoch {})"
|
45 |
+
.format(checkpoint_path, checkpoint['epoch']))
|
46 |
+
```
|