Update ocr_engine.py

ocr_engine.py

@@ -4,39 +4,37 @@ import cv2
 import numpy as np
 from PIL import Image
 
-# Initialize EasyOCR reader (only once)
+# Initialize EasyOCR reader
 reader = easyocr.Reader(['en'], gpu=False)
 
 def preprocess_image(image):
     # Convert to grayscale
     gray = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
 
-    # Apply thresholding (adaptive works well for 7-seg)
-    thresh = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_MEAN_C,
-                                   cv2.THRESH_BINARY_INV, 15, 10)
+    # Blur + Otsu thresholding
+    blur = cv2.GaussianBlur(gray, (3, 3), 0)
+    _, thresh = cv2.threshold(blur, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
 
-    # Dilation to strengthen numbers
-    kernel = np.ones((2, 2), np.uint8)
-    dilated = cv2.dilate(thresh, kernel, iterations=1)
-
-    return dilated
+    return thresh
 
 def extract_weight_from_image(pil_image):
     try:
-        # Convert PIL to OpenCV
+        # Convert PIL image to OpenCV image
         image = np.array(pil_image.convert("RGB"))
         processed = preprocess_image(image)
 
         # OCR
         result = reader.readtext(processed)
+        print("OCR Results:", result)  # for debugging
 
-        # Filter and extract digits like weight (e.g., 75.5)
         weight = None
         confidence = 0.0
+
         for detection in result:
             text = detection[1]
             conf = detection[2]
-            match = re.search(r"\d{2,4}(\.\d{1,2})?", text)  # match 2-4 digit decimal
+
+            match = re.search(r"\b\d+(\.\d+)?\b", text)  # more flexible matching
             if match:
                 weight = match.group()
                 confidence = conf