Update ocr_engine.py

ocr_engine.py

@@ -2,30 +2,30 @@ from transformers import TrOCRProcessor, VisionEncoderDecoderModel
 from PIL import Image, ImageEnhance
 import re
 
-# Load TrOCR model
-processor = TrOCRProcessor.from_pretrained("microsoft/trocr-base-handwritten")
-model = VisionEncoderDecoderModel.from_pretrained("microsoft/trocr-base-handwritten")
+# ✅ Load printed text model (more accurate for digital fonts)
+processor = TrOCRProcessor.from_pretrained("microsoft/trocr-base-printed")
+model = VisionEncoderDecoderModel.from_pretrained("microsoft/trocr-base-printed")
 
 def extract_weight(image: Image.Image) -> str:
-    # Step 1: Preprocess image
+    # Preprocess: convert to grayscale and enhance contrast
     image = image.convert("L")  # grayscale
     image = ImageEnhance.Contrast(image).enhance(2.0)
     image = ImageEnhance.Sharpness(image).enhance(2.5)
     image = image.convert("RGB")
 
-    # Step 2: Run Hugging Face OCR
+    # Run OCR
     pixel_values = processor(images=image, return_tensors="pt").pixel_values
     generated_ids = model.generate(pixel_values, max_length=32)
     full_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
 
     print("OCR Output:", full_text)
 
-    # Step 3: Extract numeric weight
+    # Extract number
     cleaned = full_text.lower().replace(" ", "")
     match = re.search(r"(\d+(\.\d+)?)", cleaned)
     weight = match.group(1) if match else None
 
-    # Step 4: Decide unit
+    # Determine unit
     if any(u in cleaned for u in ["kg", "kgs", "kilo"]):
         unit = "kg"
     elif any(u in cleaned for u in ["g", "gram", "grams"]):