Spaces:
Runtime error
Runtime error
danyalmalik
commited on
Commit
•
5dfae46
1
Parent(s):
e247352
switched to transfer learning
Browse files
app.py
CHANGED
@@ -1,27 +1,29 @@
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
-
|
|
|
|
|
4 |
import numpy as np
|
5 |
import os
|
6 |
import huggingface_hub
|
7 |
|
8 |
-
from net import Net
|
9 |
-
|
10 |
if torch.cuda.is_available():
|
11 |
device = torch.device("cuda:0")
|
12 |
else:
|
13 |
device = torch.device("cpu")
|
14 |
|
15 |
|
16 |
-
net =
|
|
|
17 |
net.to(device)
|
18 |
|
19 |
HF_Token = os.environ['HF_Token']
|
20 |
|
21 |
model = huggingface_hub.cached_download(huggingface_hub.hf_hub_url(
|
22 |
-
'danyalmalik/sceneryclassifier', '
|
23 |
|
24 |
net.load_state_dict(torch.load(model, map_location=device))
|
|
|
25 |
|
26 |
mean = np.array([0.5, 0.5, 0.5])
|
27 |
std = np.array([0.25, 0.25, 0.25])
|
@@ -49,9 +51,10 @@ def predict(img):
|
|
49 |
try:
|
50 |
img = data_transforms(img)
|
51 |
img = img.to(device)
|
|
|
52 |
|
53 |
with torch.no_grad():
|
54 |
-
output = net(img)
|
55 |
|
56 |
pred = [output[0][i].item() for i in range(len(labels))]
|
57 |
|
@@ -62,5 +65,5 @@ def predict(img):
|
|
62 |
return weightage
|
63 |
|
64 |
|
65 |
-
gr.Interface(fn=predict, inputs=gr.Image(
|
66 |
outputs='label', title=title, examples=examples()).launch()
|
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
+
import torch.nn as nn
|
4 |
+
import torch.nn.functional as F
|
5 |
+
from torchvision import transforms, models
|
6 |
import numpy as np
|
7 |
import os
|
8 |
import huggingface_hub
|
9 |
|
|
|
|
|
10 |
if torch.cuda.is_available():
|
11 |
device = torch.device("cuda:0")
|
12 |
else:
|
13 |
device = torch.device("cpu")
|
14 |
|
15 |
|
16 |
+
net = models.resnet18(pretrained=False)
|
17 |
+
net.fc = nn.Linear(net.fc.in_features, 6)
|
18 |
net.to(device)
|
19 |
|
20 |
HF_Token = os.environ['HF_Token']
|
21 |
|
22 |
model = huggingface_hub.cached_download(huggingface_hub.hf_hub_url(
|
23 |
+
'danyalmalik/sceneryclassifier', '1655988285.9725637_Acc0.88_modelweights.pth'), use_auth_token=HF_Token)
|
24 |
|
25 |
net.load_state_dict(torch.load(model, map_location=device))
|
26 |
+
net.eval()
|
27 |
|
28 |
mean = np.array([0.5, 0.5, 0.5])
|
29 |
std = np.array([0.25, 0.25, 0.25])
|
|
|
51 |
try:
|
52 |
img = data_transforms(img)
|
53 |
img = img.to(device)
|
54 |
+
img = img.unsqueeze(0)
|
55 |
|
56 |
with torch.no_grad():
|
57 |
+
output = F.softmax(net(img), dim=1)
|
58 |
|
59 |
pred = [output[0][i].item() for i in range(len(labels))]
|
60 |
|
|
|
65 |
return weightage
|
66 |
|
67 |
|
68 |
+
gr.Interface(fn=predict, inputs=gr.Image(type='pil'),
|
69 |
outputs='label', title=title, examples=examples()).launch()
|
net.py
DELETED
@@ -1,35 +0,0 @@
|
|
1 |
-
import torch
|
2 |
-
import torch.nn as nn
|
3 |
-
import torch.nn.functional as F
|
4 |
-
|
5 |
-
|
6 |
-
class Net(nn.Module):
|
7 |
-
def __init__(self):
|
8 |
-
super().__init__()
|
9 |
-
self.conv1 = nn.Conv2d(3, 32, 5)
|
10 |
-
self.conv2 = nn.Conv2d(32, 64, 5)
|
11 |
-
self.conv3 = nn.Conv2d(64, 128, 5)
|
12 |
-
|
13 |
-
x = torch.randn(3, 150, 150).view(-1, 3, 150, 150)
|
14 |
-
self._to_linear = None
|
15 |
-
self.convs(x)
|
16 |
-
|
17 |
-
self.fc1 = nn.Linear(self._to_linear, 512)
|
18 |
-
self.fc2 = nn.Linear(512, 6)
|
19 |
-
|
20 |
-
def convs(self, x):
|
21 |
-
x = F.max_pool2d(F.relu(self.conv1(x)), (2, 2))
|
22 |
-
x = F.max_pool2d(F.relu(self.conv2(x)), (2, 2))
|
23 |
-
x = F.max_pool2d(F.relu(self.conv3(x)), (2, 2))
|
24 |
-
|
25 |
-
if self._to_linear is None:
|
26 |
-
self._to_linear = x[0].shape[0]*x[0].shape[1]*x[0].shape[2]
|
27 |
-
return x
|
28 |
-
|
29 |
-
def forward(self, x):
|
30 |
-
x = self.convs(x)
|
31 |
-
x = x.view(-1, self._to_linear)
|
32 |
-
x = F.relu(self.fc1(x))
|
33 |
-
x = self.fc2(x)
|
34 |
-
|
35 |
-
return F.softmax(x, dim=1)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|