change app.py

app.py

@@ -3,13 +3,46 @@ from transformers import VisionEncoderDecoderModel, TrOCRProcessor
 from PIL import Image
 
 # Load the model and processor from Hugging Face
-model = VisionEncoderDecoderModel.from_pretrained("paudelanil/denvagari-TrOCR")
-processor = TrOCRProcessor.from_pretrained("paudelanil/denvagari-TrOCR")
+model = VisionEncoderDecoderModel.from_pretrained("paudelanil/trocr-devanagari")
+processor = TrOCRProcessor.from_pretrained("paudelanil/trocr-devanagari")
+
+device = 'cuda' if torch.cuda.is_available() else 'cpu'
+model.to(device)
+def preprocess_image(image):
+        # Resize while maintaining aspect ratio
+        target_size = (224, 224)
+        original_size = image.size
+
+        # Calculate the new size while maintaining aspect ratio
+        aspect_ratio = original_size[0] / original_size[1]
+        if aspect_ratio > 1:  # Width is greater than height
+            new_width = target_size[0]
+            new_height = int(target_size[0] / aspect_ratio)
+        else:  # Height is greater than width
+            new_height = target_size[1]
+            new_width = int(target_size[1] * aspect_ratio)
+
+        # Resize the image
+        resized_img = image.resize((new_width, new_height))
+
+        # Calculate padding values
+        padding_width = target_size[0] - new_width
+        padding_height = target_size[1] - new_height
+
+        # Apply padding to center the resized image
+        pad_left = padding_width // 2
+        pad_top = padding_height // 2
+        pad_image = Image.new('RGB', target_size, (255, 255, 255))  # White background
+        pad_image.paste(resized_img, (pad_left, pad_top))
+
+        return pad_image
+    
 
 def predict(image):
     # Preprocess the image
     image = Image.open(image).convert("RGB")
-    pixel_values = processor(image, return_tensors="pt").pixel_values
+    image = preprocess_image(image)
+    pixel_values = processor(image, return_tensors="pt").pixel_values.to(device)
     
     # Generate text from the image
     generated_ids = model.generate(pixel_values)
@@ -27,4 +60,4 @@ interface = gr.Interface(
 )
 
 # Launch the interface
-interface.launch()
+interface.launch(share=True)