Added new classifier

models/{nepali_english_classifier_n.h5 → dev_roman_classifier.pth}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ea47b670f323d058fc94b2c8618dec19fd45f481b0b9a67cd22cdf03dad938cf
-size 97901776
+oid sha256:54be1436e92b8e32bfa6f62b59a5af27181999ae8f9e08452f1fe699e8b6fef7
+size 95411068

models/{label_encoder_n.pkl → dev_roman_label_encoder.pkl}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0f13968e5cc1ea650cda347be74bf3ff5816ea4c3b6ded778ce05fe7863f44fe
-size 545
+oid sha256:9ca7716f756b0397c777fa79fa61d6e64c174fe03c598c97b50889e0c587700f
+size 298

utils.py

@@ -374,24 +374,27 @@ def predict_ne(image_path, device="cpu"):
         # load label encoder
         
         device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
-        model = ResNetClassifier(num_classes=2).to(device)
+        model = ResNetClassifier(num_classes=4).to(device)
         # model.eval()
         transform = transforms.Compose([
-        transforms.Resize((224, 224)),
+        transforms.Resize(256),                      # Resize shorter side to 256
+        transforms.CenterCrop(224),                  # Crop center 224x224 patch
         transforms.ToTensor(),
-        transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
-    ])
+        transforms.Normalize(mean=[0.485, 0.456, 0.406],
+                            std=[0.229, 0.224, 0.225])
+            ])
+
         image = Image.open(image_path).convert('RGB')
         image_tensor = transform(image).unsqueeze(0).to(device)
-        model.load_state_dict(torch.load('models/nepali_english_classifier.pth', map_location=device))
+        
+        # loading model weights/state_dict
+        model.load_state_dict(torch.load('models/dev_roman_classifier.pth', map_location=device))
         model.eval()
+
+        # loading label encoder
+        with open('models/dev_roman_label_encoder.pkl', 'rb') as f:
+            le = pickle.load(f)
         with torch.no_grad():
             output = model(image_tensor)
             _, predicted = torch.max(output, 1)
-        print(predicted.item())
-        if predicted.item() == 0:
-            return "Nepali"
-        elif predicted.item() == 1:
-            return "English"
-        else:
-         return predicted.item()
+        return le.inverse_transform([predicted.item()])[0]