Spaces:
Runtime error
Runtime error
Upload Models.py
Browse files
Models.py
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import torch
|
2 |
+
import torch.nn as nn
|
3 |
+
import torch.optim as optim
|
4 |
+
import torch.nn.functional as F
|
5 |
+
import timm
|
6 |
+
|
7 |
+
# ResNet50 Model
|
8 |
+
class ResNet(nn.Module):
|
9 |
+
def __init__(self, num_classes, is_freeze=True):
|
10 |
+
super(ResNet, self).__init__()
|
11 |
+
|
12 |
+
self.num_classes = num_classes
|
13 |
+
self.is_freeze = is_freeze
|
14 |
+
self.base_model = timm.create_model('resnet50', pretrained=True)
|
15 |
+
|
16 |
+
if self.is_freeze:
|
17 |
+
for param in self.base_model.parameters():
|
18 |
+
param.requires_grad = False
|
19 |
+
|
20 |
+
self.base_model.fc = nn.Linear(2048, self.num_classes)
|
21 |
+
|
22 |
+
def forward(self, x):
|
23 |
+
x = self.base_model(x)
|
24 |
+
return x
|
25 |
+
|
26 |
+
# EfficientNet Model
|
27 |
+
class EfficientNet(nn.Module):
|
28 |
+
def __init__(self, num_classes):
|
29 |
+
super(EfficientNet, self).__init__()
|
30 |
+
|
31 |
+
self.num_classes = num_classes
|
32 |
+
self.base_model = timm.create_model('efficientnet_b0', pretrained=True)
|
33 |
+
self.base_model.classifier = nn.Linear(1280, self.num_classes)
|
34 |
+
|
35 |
+
def forward(self, x):
|
36 |
+
x = self.base_model(x)
|
37 |
+
return x
|
38 |
+
|
39 |
+
# BaseLine Model
|
40 |
+
class BaseLine(nn.Module):
|
41 |
+
def __init__(self, num_classes):
|
42 |
+
super(BaseLine, self).__init__()
|
43 |
+
|
44 |
+
self.Conv1 = nn.Conv2d(3, 96, kernel_size=11, stride=4, padding=1)
|
45 |
+
self.Conv2 = nn.Conv2d(96, 256, kernel_size=5, padding=2)
|
46 |
+
self.Conv3 = nn.Conv2d(256, 384, kernel_size=3, padding=1)
|
47 |
+
self.Conv4 = nn.Conv2d(384, 256, kernel_size=3, padding=1)
|
48 |
+
|
49 |
+
self.Linear1 = nn.Linear(2304, 512)
|
50 |
+
self.Linear3 = nn.Linear(512, num_classes)
|
51 |
+
|
52 |
+
self.relu = nn.ReLU()
|
53 |
+
self.dropout = nn.Dropout(p=0.5)
|
54 |
+
|
55 |
+
self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2)
|
56 |
+
self.flatten = nn.Flatten()
|
57 |
+
|
58 |
+
def forward(self, x):
|
59 |
+
x = self.Conv1(x)
|
60 |
+
x = self.relu(x)
|
61 |
+
x = self.maxpool(x)
|
62 |
+
|
63 |
+
x = self.Conv2(x)
|
64 |
+
x = self.maxpool(x)
|
65 |
+
|
66 |
+
x = self.Conv3(x)
|
67 |
+
x = self.Conv4(x)
|
68 |
+
x = self.maxpool(x)
|
69 |
+
|
70 |
+
x = self.flatten(x)
|
71 |
+
x = self.Linear1(x)
|
72 |
+
x = self.relu(x)
|
73 |
+
x = self.dropout(x)
|
74 |
+
|
75 |
+
x = self.Linear3(x)
|
76 |
+
return x
|