update app.py

app.py

@@ -1,31 +1,38 @@
 
 import torch
 from torch import nn
-from torchvision import datasets
+import torch.nn.functional as F
 from torchvision.transforms import ToTensor
 
-
 # Define model
-class NeuralNetwork(nn.Module):
+class ConvNet(nn.Module):
     def __init__(self):
-        super().__init__()
-        self.flatten = nn.Flatten()
-        self.linear_relu_stack = nn.Sequential(
-            nn.Linear(28*28, 512),
-            nn.ReLU(),
-            nn.Linear(512, 512),
-            nn.ReLU(),
-            nn.Linear(512, 10)
-        )
+        super(ConvNet, self).__init__()
+        self.conv1 = nn.Conv2d(1, 32, kernel_size=5)
+        self.conv2 = nn.Conv2d(32, 32, kernel_size=5)
+        self.conv3 = nn.Conv2d(32,64, kernel_size=5)
+        self.fc1 = nn.Linear(3*3*64, 256)
+        self.fc2 = nn.Linear(256, 10)
 
     def forward(self, x):
-        x = self.flatten(x)
-        logits = self.linear_relu_stack(x)
+        x = F.relu(self.conv1(x))
+        #x = F.dropout(x, p=0.5, training=self.training)
+        x = F.relu(F.max_pool2d(self.conv2(x), 2))
+        x = F.dropout(x, p=0.5, training=self.training)
+        x = F.relu(F.max_pool2d(self.conv3(x),2))
+        x = F.dropout(x, p=0.5, training=self.training)
+        x = x.view(-1,3*3*64 )
+        x = F.relu(self.fc1(x))
+        x = F.dropout(x, training=self.training)
+        logits = self.fc2(x)
         return logits
 
 
-model = NeuralNetwork()
-model.load_state_dict(torch.load("model_mnist_mlp.pth"))
+model = ConvNet()
+model.load_state_dict(
+    torch.load("weights/mnist_convnet_model.pth",
+    map_location=torch.device('cpu'))
+    )
 
 model.eval()
 
@@ -40,25 +47,24 @@ def predict(image):
         prob = torch.nn.functional.softmax(pred[0], dim=0)
 
     confidences = {i: float(prob[i]) for i in range(10)} 
-
     return confidences
 
 
-with gr.Blocks(css=".gradio-container {background:lightyellow;color:red;}", title="テスト"
+with gr.Blocks(css=".gradio-container {background:honeydew;}", title="MNIST 分類器"
                ) as demo:
-    gr.HTML('<div style="font-size:12pt; text-align:center; color:yellow;"MNIST 分類器</div>')
+    gr.HTML("""<div style="font-family:'Times New Roman', 'Serif'; font-size:16pt; font-weight:bold; text-align:center; color:royalblue;">MNIST 分類器</div>""")
 
     with gr.Row():
         with gr.Tab("キャンバス"):
             input_image1 = gr.Image(label="画像入力", source="canvas", type="pil", image_mode="L", shape=(28,28), invert_colors=True)    
-            send_btn1 = gr.Button("予測する")
+            send_btn1 = gr.Button("推論する")
 
         with gr.Tab("画像ファイル"):
             input_image2 = gr.Image(label="画像入力", type="pil", image_mode="L", shape=(28, 28), invert_colors=True)
-            send_btn2 = gr.Button("予測する")
-            gr.Examples(['examples/example02.png', 'examples/example04.png'], inputs=input_image2)
+            send_btn2 = gr.Button("推論する")
+            gr.Examples(['examples/sample02.png', 'examples/sample04.png'], inputs=input_image2)
 
-        output_label=gr.Label(label="予測確率", num_top_classes=5)
+        output_label=gr.Label(label="推論確率", num_top_classes=3)
 
     send_btn1.click(fn=predict, inputs=input_image1, outputs=output_label)
     send_btn2.click(fn=predict, inputs=input_image2, outputs=output_label)
@@ -66,4 +72,5 @@ with gr.Blocks(css=".gradio-container {background:lightyellow;color:red;}", titl
 # demo.queue(concurrency_count=3)
 demo.launch()
 
+
 ### EOF ###