update

models/model.py

@@ -0,0 +1,28 @@
+# models/model.py
+import torch
+import torch.nn as nn
+import torchvision.models as models
+
+# Deep Learning Model
+class DeepLearningModel(nn.Module):
+    def __init__(self):
+        super().__init__()
+        model = models.vgg16(pretrained=True)
+        self.features = model.features
+        self.avgpool = model.avgpool
+        self.classifier = model.classifier
+        # Modify the classifier for 3 classes
+        self.classifier[6] = nn.Sequential(
+            nn.Linear(4096, 512),
+            nn.BatchNorm1d(512),  
+            nn.ReLU(),
+            nn.Dropout(0.5),
+            nn.Linear(512, 3)
+        )
+
+    def forward(self, x):
+        x = self.features(x)
+        x = self.avgpool(x)
+        x = torch.flatten(x, 1)
+        x = self.classifier(x)
+        return x

models/vgg16_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76230eaadac39b356db06bc647089ce09ad04106a4641765cf8a30bfd1b6baf3
+size 545461325

requirements.txt

@@ -0,0 +1,8 @@
+streamlit==1.26.0
+torch==2.0.1
+torchvision==0.15.2
+numpy==1.23.5
+Pillow<10
+opencv-python==4.8.0.76
+joblib==1.3.2
+grad-cam

scripts/xai_eval.py

@@ -0,0 +1,33 @@
+from torchvision import models, transforms
+import torch.nn as nn
+from pytorch_grad_cam import GradCAM
+import torch
+
+def load_model(model_path: str):
+    """Load a fine-tuned VGG model from model path"""
+    vgg_model = models.vgg16(pretrained=False)
+    vgg_model.classifier[6] = nn.Sequential(
+        nn.Linear(vgg_model.classifier[6].in_features, 512),
+        nn.BatchNorm1d(512),
+        nn.ReLU(),
+        nn.Dropout(0.5),
+        nn.Linear(512,3)
+    ) 
+    vgg_model.load_state_dict(torch.load(model_path, map_location='cpu'))
+    vgg_model.eval()
+    return vgg_model
+
+def convert_to_gradcam(model):
+    """Initialize a Grad-CAM explainer for the provided model"""
+    target_layers = [model.features[-1]]
+    return GradCAM(model=model, target_layers=target_layers)
+
+def preprocess_image(image):
+    """Apply image pre-processing for VGG-16 model"""
+    transform = transforms.Compose([ 
+        transforms.Resize((224, 224)),
+        transforms.ToTensor(),
+        transforms.Normalize(mean=[0.3205, 0.2244, 0.1613], 
+                           std=[0.2996, 0.2158, 0.1711])
+    ])
+    return transform(image)