Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -13,23 +13,32 @@ from PIL import Image
|
|
13 |
class Net(nn.Module):
|
14 |
def __init__(self):
|
15 |
super(Net, self).__init__()
|
|
|
16 |
self.fc1 = nn.Linear(28 * 28, 128)
|
17 |
self.fc2 = nn.Linear(128, 64)
|
18 |
self.fc3 = nn.Linear(64, 10)
|
19 |
|
20 |
def forward(self, x):
|
21 |
-
|
|
|
|
|
22 |
x = F.relu(self.fc1(x))
|
23 |
x = F.relu(self.fc2(x))
|
|
|
24 |
x = self.fc3(x)
|
25 |
return F.log_softmax(x, dim=1)
|
26 |
|
27 |
# Load and preprocess the MNIST dataset
|
28 |
-
transform = transforms.Compose([
|
|
|
|
|
|
|
29 |
|
|
|
30 |
train_dataset = datasets.MNIST(root='./data', train=True, download=True, transform=transform)
|
31 |
train_loader = DataLoader(train_dataset, batch_size=32, shuffle=True)
|
32 |
|
|
|
33 |
test_dataset = datasets.MNIST(root='./data', train=False, download=True, transform=transform)
|
34 |
test_loader = DataLoader(test_dataset, batch_size=32, shuffle=False)
|
35 |
|
@@ -38,8 +47,10 @@ model = Net()
|
|
38 |
criterion = nn.CrossEntropyLoss()
|
39 |
optimizer = optim.Adam(model.parameters(), lr=0.001)
|
40 |
|
41 |
-
#
|
42 |
-
model_path = 'mnist_model.pth'
|
|
|
|
|
43 |
if not os.path.isfile(model_path):
|
44 |
raise FileNotFoundError(f"The model file '{model_path}' was not found.")
|
45 |
|
@@ -49,29 +60,29 @@ model.eval()
|
|
49 |
|
50 |
# Define the predict function
|
51 |
def predict_image(img):
|
52 |
-
# Preprocess the image
|
53 |
-
img = img.convert('L')
|
54 |
-
img = img.resize((28, 28))
|
55 |
-
img = np.array(img).astype('float32') / 255.0
|
56 |
-
img = (img - 0.5) / 0.5 # Normalize
|
57 |
img = torch.tensor(img).unsqueeze(0).unsqueeze(0) # Add batch and channel dimensions
|
58 |
|
59 |
# Make a prediction
|
60 |
with torch.no_grad():
|
61 |
-
output = model(img)
|
62 |
-
predicted_digit = output.argmax(dim=1, keepdim=True).item()
|
63 |
|
64 |
return predicted_digit
|
65 |
|
66 |
# Create the Gradio interface
|
67 |
iface = gr.Interface(
|
68 |
-
fn=predict_image,
|
69 |
-
inputs=gr.inputs.Image(shape=(28, 28), image_mode='L', invert_colors=False),
|
70 |
-
outputs='label',
|
71 |
-
live=True,
|
72 |
-
description="Upload an image of a handwritten digit, and the model will predict the digit."
|
73 |
)
|
74 |
|
75 |
-
# Launch the interface
|
76 |
if __name__ == '__main__':
|
77 |
iface.launch()
|
|
|
13 |
class Net(nn.Module):
|
14 |
def __init__(self):
|
15 |
super(Net, self).__init__()
|
16 |
+
# Define layers of the neural network
|
17 |
self.fc1 = nn.Linear(28 * 28, 128)
|
18 |
self.fc2 = nn.Linear(128, 64)
|
19 |
self.fc3 = nn.Linear(64, 10)
|
20 |
|
21 |
def forward(self, x):
|
22 |
+
# Flatten the input tensor
|
23 |
+
x = x.view(-1, 28 * 28)
|
24 |
+
# Apply ReLU activation function
|
25 |
x = F.relu(self.fc1(x))
|
26 |
x = F.relu(self.fc2(x))
|
27 |
+
# Output layer with log softmax activation
|
28 |
x = self.fc3(x)
|
29 |
return F.log_softmax(x, dim=1)
|
30 |
|
31 |
# Load and preprocess the MNIST dataset
|
32 |
+
transform = transforms.Compose([
|
33 |
+
transforms.ToTensor(),
|
34 |
+
transforms.Normalize((0.5,), (0.5,))
|
35 |
+
])
|
36 |
|
37 |
+
# Download and load training dataset
|
38 |
train_dataset = datasets.MNIST(root='./data', train=True, download=True, transform=transform)
|
39 |
train_loader = DataLoader(train_dataset, batch_size=32, shuffle=True)
|
40 |
|
41 |
+
# Download and load test dataset
|
42 |
test_dataset = datasets.MNIST(root='./data', train=False, download=True, transform=transform)
|
43 |
test_loader = DataLoader(test_dataset, batch_size=32, shuffle=False)
|
44 |
|
|
|
47 |
criterion = nn.CrossEntropyLoss()
|
48 |
optimizer = optim.Adam(model.parameters(), lr=0.001)
|
49 |
|
50 |
+
# Path to the model file
|
51 |
+
model_path = 'mnist_model.pth'
|
52 |
+
|
53 |
+
# Check if the model file exists
|
54 |
if not os.path.isfile(model_path):
|
55 |
raise FileNotFoundError(f"The model file '{model_path}' was not found.")
|
56 |
|
|
|
60 |
|
61 |
# Define the predict function
|
62 |
def predict_image(img):
|
63 |
+
# Preprocess the uploaded image
|
64 |
+
img = img.convert('L') # Convert image to grayscale
|
65 |
+
img = img.resize((28, 28)) # Resize image to 28x28 pixels
|
66 |
+
img = np.array(img).astype('float32') / 255.0 # Normalize pixel values
|
67 |
+
img = (img - 0.5) / 0.5 # Normalize to range [-1, 1]
|
68 |
img = torch.tensor(img).unsqueeze(0).unsqueeze(0) # Add batch and channel dimensions
|
69 |
|
70 |
# Make a prediction
|
71 |
with torch.no_grad():
|
72 |
+
output = model(img) # Forward pass through the model
|
73 |
+
predicted_digit = output.argmax(dim=1, keepdim=True).item() # Get the predicted digit
|
74 |
|
75 |
return predicted_digit
|
76 |
|
77 |
# Create the Gradio interface
|
78 |
iface = gr.Interface(
|
79 |
+
fn=predict_image, # Function to be called on image upload
|
80 |
+
inputs=gr.inputs.Image(shape=(28, 28), image_mode='L', invert_colors=False), # Input format
|
81 |
+
outputs='label', # Output format
|
82 |
+
live=True, # Live update
|
83 |
+
description="Upload an image of a handwritten digit, and the model will predict the digit." # Description of the interface
|
84 |
)
|
85 |
|
86 |
+
# Launch the Gradio interface
|
87 |
if __name__ == '__main__':
|
88 |
iface.launch()
|