Update app.py

app.py

@@ -35,7 +35,7 @@ inv_normalize = T.Normalize(
 
 grad_cams = [GradCAM(model=model, target_layers=[model.convblock3[i]], use_cuda=False) for i in range(5)]
 
-def get_gradcam_image(input_tensor, label, target_layer):
+def create_gradcam(input_tensor, label, target_layer):
     grad_cam = grad_cams[target_layer]
     targets = [ClassifierOutputTarget(label)]
     grayscale_cam = grad_cam(input_tensor=input_tensor, targets=targets)
@@ -43,7 +43,7 @@ def get_gradcam_image(input_tensor, label, target_layer):
     return grayscale_cam
 
 
-def image_classifier(input_image, top_classes=3, show_cam=True, target_layers=[2, 3], transparency=0.5):
+def inference(input_image, top_classes=3, show_cam=True, target_layers=[2, 3], transparency=0.5):
     orig_image = input_image
     input_image = transform(input_image)
      
@@ -54,33 +54,33 @@ def image_classifier(input_image, top_classes=3, show_cam=True, target_layers=[2
     o = softmax(output.flatten())
 
     confidences = {classes[i]: float(o[i]) for i in range(10)}
-    confidences_sorted = dict(sorted(confidences.items(), key=lambda x:x[1],reverse=True))
-    confidences = {k: confidences_sorted[k] for k in list(confidences_sorted)[:top_classes]}
+    confidences = dict(sorted(confidences.items(), key=lambda x:x[1],reverse=True))
+    confidences = {i: confidences[i] for i in list(confidences)[:top_classes]}
     _, label = torch.max(output, 1)
     
     outputs = list()
     if show_cam:
         for layer in target_layers:
-            grayscale_cam = get_gradcam_image(input_image, label, layer)
+            grayscale_cam = create_gradcam(input_image, label, layer)
             output_image = show_cam_on_image(orig_image / 255, grayscale_cam, use_rgb=True, image_weight=transparency)
             outputs.append((output_image, f"Layer {layer - 5}"))    
 
     return outputs, confidences
 
-#examples = [["examples/cat.jpg", 3, True,["-2","-1"],0.5], ["examples/dog.jpg", 3, True,["-2","-1"],0.5]]
+
 examples = []
 for i in range(10):
   examples.append([f'examples/{classes[i]}.jpg', 3, True,["-2","-1"],0.5])
 
 demo_1 = gr.Interface(
-    fn=image_classifier,
+    fn=inference,
     inputs=[
         gr.Image(shape=(32, 32), label="Input Image").style(width=128, height=128),
-        gr.Slider(1, 10, value=3, step=1, label="Top Classes",
+        gr.Slider(1, 10, value=3, step=1, label="Top Predictions",
                   info="How many top classes do you want to see?"),
-        gr.Checkbox(label="Enable GradCAM", value=True, info="Do you want to see GradCAM Images?"),
-        gr.CheckboxGroup(["-5","-4", "-3", "-2", "-1"], value=["-2", "-1"], label="Network Layers", type='index',
-                         info="Which layer(s) GradCAM do you want to visualize?",),
+        gr.Checkbox(label="Show GradCAM", value=True, info="Do you want to see GradCAM Images?"),
+        gr.CheckboxGroup(["-5","-4", "-3", "-2", "-1"], value=["-2", "-1"], label="Conv Layers", type='index',
+                         info="For which layer do you want to visualize GradCAM?",),
         gr.Slider(0, 1, value=0.5, label="Transparency", step=0.1,
                   info="Set Transparency of CAMs")
     ],
@@ -89,7 +89,7 @@ demo_1 = gr.Interface(
 )
 
 
-def show_incorrect(num_examples=10):
+def show_misclassified(num_examples=10):
     result = list()
     for i in range(num_examples):
         j = np.random.randint(1,30)
@@ -97,19 +97,19 @@ def show_incorrect(num_examples=10):
         actual = classes[wrong_img.loc[j-1].at["actual"]]
         predicted = classes[wrong_img.loc[j-1].at["predicted"]]
         
-        result.append((image, f"Actual:{actual} / Predicted:{predicted}"))
+        result.append((image, f"Actual:{actual} \\n / Predicted:{predicted}"))
         
     return result
 
 
 demo_2 = gr.Interface(
-    fn=show_incorrect,
+    fn=show_misclassified,
     inputs=[
-        gr.Number(value=10, minimum=1, maximum=30, label="Input number(s) of images", precision=0,
+        gr.Number(value=10, minimum=1, maximum=30, label="Input number of images", precision=0,
                   info="How many misclassified examples do you want to view? (max 30)")
     ],
     outputs=[gr.Gallery(label="Misclassified Images (Actual / Predicted)", columns=5)]
 )
 
-demo = gr.TabbedInterface([demo_1, demo_2], ["Image Classifier", "Misclassified Images"])
+demo = gr.TabbedInterface([demo_1, demo_2], ["CIFAR10 Classifier", "Mis-predicted Images"])
 demo.launch(debug=True)