first commit

mnist_app/app.py

@@ -0,0 +1,28 @@
+from model import load_model
+
+import gradio as gr
+import os
+import torch
+
+
+
+model,transforming,classes = load_model()
+
+def predict(img):
+    img = transforming(img)
+    img = img.unsqueeze(0)
+
+
+    model.eval()
+    with torch.inference_mode():
+        pred_probs = torch.softmax(model(img), dim=1)
+    return {str(i): float(pred_probs[0][i]) for i in range(len(pred_probs[0]))}
+
+title = 'MNIST Digit Prediction'
+description = 'Predict handwritten digits (0-9) using a trained model.'
+inputs = gr.Image(type='pil', label='Upload an image of a digit')
+outputs = gr.Label(num_top_classes=3, label='Predictions')
+demo = gr.Interface(fn=predict, inputs=inputs, outputs=outputs, title=title, description=description)
+
+# Launch the interface
+demo.launch()

mnist_app/best_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6f4516312e2020aa20cde939bfa5663d3001cfe79ebd70ef3ab75f495f05298
+size 24612

mnist_app/model.py

@@ -0,0 +1,55 @@
+
+import torch
+from torch import nn
+from torchvision import transforms
+
+
+class MnistModel(nn.Module):
+
+    classes = ['0 - zero',
+                '1 - one',
+                '2 - two',
+                '3 - three',
+                '4 - four',
+                '5 - five',
+                '6 - six',
+                '7 - seven',
+                '8 - eight',
+                '9 - nine']
+    def __init__(self, *args, **kwargs) -> None:
+        super().__init__(*args, **kwargs)
+        self.conv1 = nn.Conv2d(1, 3, 3)
+        self.conv2 = nn.Conv2d(3, 6, 3)
+        self.maxpool = nn.MaxPool2d(2, 2)
+        self.fc1 = nn.Linear(150, 32)
+        self.fc2 = nn.Linear(32, 10)
+        #self.fc3 = nn.Linear(32, 10)
+        self.dropout = nn.Dropout(0.3) 
+
+    def forward(self, x):
+        l1 = nn.ReLU()(self.conv1(x))
+        l1 = self.maxpool(l1)
+        l2 = nn.ReLU()(self.conv2(l1))
+        l2 = self.maxpool(l2)
+        fc = torch.flatten(l2, 1)
+        fc1 = nn.ReLU()(self.fc1(fc))
+        fc1 = self.dropout(fc1)  
+        #fc2 = nn.ReLU()(self.fc2(fc1))
+        out = self.fc2(fc1)
+        return out
+    
+
+def load_model():
+    model = MnistModel()
+    transforming = transforms.Compose([
+    transforms.Resize((28,28)),
+    transforms.ToTensor(),
+    transforms.Grayscale(num_output_channels=1)
+])
+    
+    model.load_state_dict(torch.load('demos/mnist_app/best_model.pth',map_location='cpu'))
+
+    return model,transforming,model.classes
+
+if __name__=='__main__':
+    pass

mnist_app/requirements.txt

@@ -0,0 +1,5 @@
+gradio==4.31.5
+pathlib==1.0.1
+torch==2.2.2
+torchvision==0.17.2
+