Update ocr_engine.py

ocr_engine.py

@@ -5,18 +5,29 @@ import re
 reader = easyocr.Reader(['en'], gpu=False)
 
 def extract_weight_from_image(img_cv):
-    # Resize to help with clarity
+    # Resize for clarity
     img = cv2.resize(img_cv, None, fx=2.0, fy=2.0, interpolation=cv2.INTER_CUBIC)
 
-    # Convert to grayscale if needed
+    # Convert to grayscale
     if len(img.shape) == 3:
         img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
 
+    # Adaptive threshold to boost contrast
+    img = cv2.adaptiveThreshold(img, 255, cv2.ADAPTIVE_THRESH_MEAN_C, 
+                                cv2.THRESH_BINARY, 11, 3)
+
+    # Morphological close to unify digits
+    kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (2, 2))
+    img = cv2.morphologyEx(img, cv2.MORPH_CLOSE, kernel)
+
+    # OCR detection
     results = reader.readtext(img)
 
-    raw_text = " ".join([text for _, text, _ in results])
-    matches = re.findall(r'\d+\.\d+|\d+', raw_text)
+    # Filter only short, confident numeric candidates
+    filtered = [text for _, text, conf in results if len(text.strip()) <= 5 and conf > 0.4]
+    raw_text = " ".join(filtered)
 
+    matches = re.findall(r"\d+\.\d+|\d+", raw_text)
     weight = matches[0] if matches else "0.0"
     confidence = int(results[0][2] * 100) if results else 0