trying still.....

app.py

@@ -8,23 +8,36 @@ from torchvision import transforms
 # --------------------------
 # Make sure this matches the architecture you used to train your model
 
+
 from torchvision.models import resnet50
+import torch.nn as nn
 
-class MyCarClassifier(torch.nn.Module):
+class MyCarClassifier(nn.Module):
     def __init__(self, num_classes=196):
         super(MyCarClassifier, self).__init__()
-        self.model = resnet50(pretrained=False)
-        in_ch = self.model.fc.in_features
-        self.model.fc = torch.nn.Linear(in_ch, num_classes)
+        self.backbone = resnet50(weights=None)
+        in_ch = self.backbone.fc.in_features
+        self.backbone.fc = nn.Identity()
+        self.fc = nn.Sequential(
+            nn.Linear(in_ch, 512),   # fc.1
+            nn.ReLU(),
+            nn.BatchNorm1d(512),     # fc.3
+            nn.Linear(512, 256),     # fc.5
+            nn.ReLU(),
+            nn.BatchNorm1d(256),     # fc.7
+            nn.Linear(256, num_classes) # fc.9
+        )
+
     def forward(self, x):
-        return self.model(x)
+        x = self.backbone(x)
+        x = self.fc(x)
+        return x
 
 # --------------------------
 # 2. Load model weights
 # --------------------------
 model = MyCarClassifier()
-state_dict = torch.load("best_stanford_cars_transfer_model.pth", map_location="cpu")
-model.model.load_state_dict(state_dict, strict=True)
+model.load_state_dict(torch.load("best_stanford_cars_transfer_model.pth", map_location="cpu"), strict=True)
 model.eval()  # important for inference
 
 # --------------------------