Update app.py

app.py

@@ -5,12 +5,14 @@ import gradio as gr
 from PIL import Image
 from pytorch_grad_cam import GradCAM
 from pytorch_grad_cam.utils.image import show_cam_on_image
-from model import CustomResNet
+from resnet import custom_ResNet
+import gradio as gr
+import os
+
 
-PATH = "model.pth"
-model= CustomResNet()
-model.load_state_dict(torch.load("model.pth", map_location=torch.device('cpu')), strict=False)
-#model.to(device)
+model = custom_ResNet()
+model.load_state_dict(torch.load("custom_resnet_model.pth", map_location=torch.device('cpu')), strict=False)
+model.setup(stage="test")
 
 inv_normalize = transforms.Normalize(
     mean=[-0.50/0.23, -0.50/0.23, -0.50/0.23],
@@ -19,18 +21,18 @@ inv_normalize = transforms.Normalize(
 classes = ('plane', 'car', 'bird', 'cat', 'deer',
            'dog', 'frog', 'horse', 'ship', 'truck')
 
-def inference(input_img, transparency = 0.5, target_layer_number = -1):
+
+def inference(input_img, transparency=0.5, target_layer_number=-1, top_classes=3):
     transform = transforms.ToTensor()
     org_img = input_img
     input_img = transform(input_img)
-    input_img = input_img
     input_img = input_img.unsqueeze(0)
     outputs = model(input_img)
-    softmax = torch.nn.Softmax(dim=0)
-    o = softmax(outputs.flatten())
-    confidences = {classes[i]: float(o[i]) for i in range(10)}
+    softmax = torch.nn.Softmax(dim=1)  # Use dim=1 to compute softmax along the classes dimension
+    o = softmax(outputs)
+    confidences = {classes[i]: float(o[0, i]) for i in range(10)}
     _, prediction = torch.max(outputs, 1)
-    target_layers = [model.layer2[target_layer_number]]
+    target_layers = [model.convblock2_l1]
     cam = GradCAM(model=model, target_layers=target_layers, use_cuda=False)
     grayscale_cam = cam(input_tensor=input_img, targets=None)
     grayscale_cam = grayscale_cam[0, :]
@@ -38,33 +40,67 @@ def inference(input_img, transparency = 0.5, target_layer_number = -1):
     img = inv_normalize(img)
     rgb_img = np.transpose(img, (1, 2, 0))
     rgb_img = rgb_img.numpy()
-    visualization = show_cam_on_image(org_img/255, grayscale_cam, use_rgb=True, image_weight=transparency)
-    return confidences, visualization
-
-title = "CIFAR10 trained on CustomBlock Model with GradCAM"
-description = "A simple Gradio interface to infer on ResNet model, and get GradCAM results"
-examples = [["cat.jpg", 0.5, -1], 
-            ["dog.jpg", 0.5, -1],
-            ["plane.jpg", 0.5, -1],
-            ["car.jpg", 0.5, -1],
-            ["bird.jpg", 0.5, -1],
-            ["deer.jpg", 0.5, -1],
-            ["frog.jpg", 0.5, -1],
-            ["horse.jpg", 0.5, -1],
-            ["ship.jpg", 0.5, -1],
-            ["truck.jpg", 0.5, -1]
-            ]
-demo = gr.Interface(
-    inference, 
-    inputs = [gr.Image(shape=(32, 32), label="Input Image"), 
-              gr.Slider(0, 1, value = 0.5, label="Opacity of GradCAM"), 
-              gr.Slider(-2, -1, value = -2, step=1, label="Which Layer?"),
-              gr.Slider(1, 10,step=1, label="How many missclassifed images"),
-              gr.Slider(1, 10,step=1, label="How many Top Class"),
-             ], 
-    outputs = [gr.Label(num_top_classes=3), gr.Image(shape=(32, 32), label="Output").style(width=128, height=128)],
-    title = title,
-    description = description,
-    examples = examples,
-)
+    visualization = show_cam_on_image(org_img / 255, grayscale_cam, use_rgb=True, image_weight=transparency)
+
+    # Sort the confidences dictionary by values in descending order
+    sorted_confidences = {k: v for k, v in sorted(confidences.items(), key=lambda item: item[1], reverse=True)}
+    # Take the top `top_classes` elements from the sorted_confidences
+    top_classes_confidences = {k: sorted_confidences[k] for k in list(sorted_confidences)[:top_classes]}
+
+    return top_classes_confidences, visualization
+
+
+# Create a wrapper function for show_misclassified_images()
+def show_misclassified_images_wrapper(num_images=10, use_gradcam=False, gradcam_layer=-1, transparency=0.5):
+    transparency = float(transparency)
+    num_images = int(num_images)
+    if use_gradcam == "Yes":
+      use_gradcam = True
+    else:
+      use_gradcam = False
+
+    return model.show_misclassified_images(num_images, use_gradcam, gradcam_layer, transparency)
+
+description1 = "Test the model's prediction. Currently the model only supports the following classes - plane, car, bird, cat, deer, dog, frog, horse, ship, truck."
+
+
+# Define the full path to the images folder
+images_folder = "examples"
+
+# Define the examples list with full paths
+examples = [[os.path.join(images_folder, "plane.jpg"), 0.5, -1],
+            [os.path.join(images_folder, "car.jpg"), 0.5, -1],
+            [os.path.join(images_folder, "bird.jpeg"), 0.5, -1],
+            [os.path.join(images_folder, "cat.jpeg"), 0.5, -1],
+            [os.path.join(images_folder, "deer.jpeg"), 0.5, -1],
+            [os.path.join(images_folder, "dog.jpeg"), 0.5, -1],
+            [os.path.join(images_folder, "frog.jpeg"), 0.5, -1],
+            [os.path.join(images_folder, "horse.jpeg"), 0.5, -1],
+            [os.path.join(images_folder, "ship.jpeg"), 0.5, -1],
+            [os.path.join(images_folder, "truck.jpeg"), 0.5, -1]]
+
+
+# Create a separate interface for the "Input an image" tab
+input_interface = gr.Interface(inference,
+                inputs=[gr.Image(shape=(32, 32), label="Input Image"),
+                        gr.Slider(0, 1, value=0.5, label="Opacity of GradCAM"),
+                        gr.Slider(-2, -1, value=-2, step=1, label="Which Layer?"),
+                        gr.Slider(1, 10, value=3, step=1, label="How many top confidence classes to be shown?")],
+                outputs=[gr.Label(),
+                         gr.Image(shape=(32, 32), label="Model Prediction").style(width=300, height=300)],
+                description=description1,examples=examples)
+
+description2 = "Displays misclassified image of the model"
+
+# Create a separate interface for the "Misclassified Images" tab
+misclassified_interface = gr.Interface(show_misclassified_images_wrapper,
+                    inputs=[gr.Number(value=10, label="Number of images to display"),
+                            gr.Radio(["Yes", "No"], value="No"  , label="Show GradCAM outputs"),
+                            gr.Slider(-2, -1, value=-1, step=1, label="Which layer for GradCAM?"),
+                            gr.Slider(0, 1, value=0.5, label="Opacity of GradCAM")],
+                    outputs=gr.Plot(), description=description2)
+
+demo = gr.TabbedInterface([input_interface, misclassified_interface], tab_names=["Input an image", "Misclassified Images"], 
+                          title="Custom Resnet on CIFAR10 using GradCAM")
+
 demo.launch()