Update app.py

app.py

@@ -1,4 +1,5 @@
 import torch
+import torchvision
 from torchvision import transforms
 import numpy as np
 import gradio as gr
@@ -11,70 +12,81 @@ inv_normalize = transforms.Normalize(
     mean=[-0.50/0.23, -0.50/0.23, -0.50/0.23],
     std=[1/0.23, 1/0.23, 1/0.23]
 )
+
 classes = ('plane', 'car', 'bird', 'cat', 'deer',
            'dog', 'frog', 'horse', 'ship', 'truck')
 
 model = LITResNet(classes)
 model.load_state_dict(torch.load("model.pth", map_location=torch.device('cpu')), strict=False)
 
-def inference(input_img, gradcam=False, num_gradcam=1, transparency=0.5, target_layer_number=-1,
-              misclassified=False, num_misclassified=1, top_classes=3):
-    input_img = np.array(input_img)
+def inference(input_img, show_gradcam, num_gradcam, layer_num, opacity, show_misclassified, num_misclassified, num_top_classes):
+    input_img = np.array(Image.fromarray(np.array(input_img)).resize((32,32)))
     org_img = input_img
-    input_img = Image.fromarray(input_img)
-    transform = transforms.Compose([
-        transforms.Resize((32, 32)),
-        transforms.ToTensor(),
-    ])
-    input_img = transform(input_img)
-    input_img = inv_normalize(input_img)
-    input_img = input_img.unsqueeze(0)
+
+    transform = transforms.ToTensor()
+    input_img = transform(input_img).unsqueeze(0)
+
     outputs = model(input_img)
-    softmax = torch.nn.Softmax(dim=1)
-    o = softmax(outputs)
-    confidences = {classes[i]: float(o[0][i]) for i in range(10)}
-    _, prediction = torch.max(outputs, 1)
-    top_indices = torch.topk(outputs, top_classes).indices.squeeze(0).tolist()
+    softmax = torch.nn.Softmax(dim=0)
+    o = softmax(outputs.flatten())
+    confidences = {classes[i]: float(o[i]) for i in range(10)}
 
-    results = [classes[idx] for idx in top_indices]
+    _, prediction = torch.max(outputs, 1)
+    is_misclassified = (prediction != labels.index(input_img_label))
 
-    if gradcam:
-        target_layers = [model.layer2[target_layer_number]]
+    if show_gradcam:
+        target_layers = [model.layer2[layer_num]]
         cam = GradCAM(model=model, target_layers=target_layers)
-        grayscale_cams = cam(input_tensor=input_img, targets=None)
-        visualizations = []
-        for i in range(num_gradcam):
-            grayscale_cam = grayscale_cams[i, :]
-            visualization = show_cam_on_image(org_img / 255, grayscale_cam, use_rgb=True, image_weight=transparency)
-            visualizations.append(visualization)
+        grayscale_cam = cam(input_tensor=input_img, targets=None)
+        grayscale_cam = grayscale_cam[0, :]
+        img = input_img.squeeze(0)
+        img = inv_normalize(img)
+        visualization = [show_cam_on_image(org_img/255, grayscale_cam, use_rgb=True, image_weight=opacity) for _ in range(num_gradcam)]
     else:
-        visualizations = None
+        visualization = []
 
-    return results, visualizations, confidences
+    if show_misclassified:
+        misclassified_imgs = [input_img for _ in range(num_misclassified)]
+    else:
+        misclassified_imgs = []
+
+    sorted_confidences = dict(sorted(confidences.items(), key=lambda item: item[1], reverse=True)[:num_top_classes])
+
+    return prediction[0].item(), classes[prediction[0].item()], is_misclassified, sorted_confidences, visualization, misclassified_imgs
 
 title = "CIFAR10 trained on ResNet18 Model with GradCAM"
-description = "A simple Gradio interface to infer on ResNet model, and get GradCAM results"
-examples = [["cat.jpg", False, 1, 0.5, -1, False, 1, 3]]
+description = "A simple Gradio interface to infer on ResNet model, get GradCAM results, and view misclassified images"
+
+examples = [
+    ["plane.jpg", True, 1, -1, 0.5, False, 0, 3],
+    ["car.jpg", True, 2, -2, 0.7, True, 1, 5],
+    ["bird.jpg", False, 0, -1, 0.5, False, 0, 3],
+    # Add more examples as needed
+]
 
 demo = gr.Interface(
     inference,
     inputs=[
         gr.Image(width=256, height=256, label="Input Image"),
-        "checkbox", 
-        gr.Number(value=1, label="Number of GradCAM Images"),
+        gr.Checkbox(value=True, label="Show GradCAM"),
+        gr.Slider(1, 5, value=1, step=1, label="Number of GradCAM Images"),
+        gr.Slider(-2, -1, value=-2, step=1, label="Which Layer?"),
         gr.Slider(0, 1, value=0.5, label="Overall Opacity of Image"),
-        gr.Slider(-1, -2, value=-1, step=1, label="Which Layer?"),
-        "checkbox",
-        gr.Number(value=1, label="Number of Misclassified Images"),
-        gr.Number(value=3, minimum=1, maximum=10, label="Top Classes to Show")
+        gr.Checkbox(value=False, label="Show Misclassified Images"),
+        gr.Slider(1, 5, value=1, step=1, label="Number of Misclassified Images"),
+        gr.Slider(1, 10, value=3, step=1, label="Number of Top Classes to Show")
     ],
     outputs=[
-        gr.Label(num_top_classes=3),
-        gr.Image(width=256, height=256, label="GradCAM Image", type="numpy"),
-        gr.Label(num_top_classes=3)
+        "text",
+        "text",
+        "text",
+        gr.Label(num_top_classes=10),
+        gr.Gallery(label="GradCAM Visualizations"),
+        gr.Gallery(label="Misclassified Images")
     ],
     title=title,
     description=description,
     examples=examples,
 )
+
 demo.launch()