Update ocr_engine.py

ocr_engine.py

@@ -5,35 +5,21 @@ from PIL import Image
 
 def extract_weight_from_image(pil_img):
     try:
-        # Step 1: Convert PIL to OpenCV
-        img = pil_img.convert("L")  # grayscale
+        # Convert PIL image to OpenCV
+        img = pil_img.convert("RGB")
         img = np.array(img)
+        img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
 
-        # Step 2: Resize image for better OCR accuracy
-        img = cv2.resize(img, None, fx=2, fy=2, interpolation=cv2.INTER_CUBIC)
+        # Preprocess
+        gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
+        blur = cv2.GaussianBlur(gray, (3, 3), 0)
 
-        # Step 3: Apply Gaussian Blur to remove noise
-        blur = cv2.GaussianBlur(img, (5, 5), 0)
-
-        # Step 4: Apply Adaptive Thresholding
-        thresh = cv2.adaptiveThreshold(
-            blur, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
-            cv2.THRESH_BINARY_INV, 11, 2
-        )
-
-        # Step 5: OCR Config - digits only
-        config = r'--oem 3 --psm 6 -c tessedit_char_whitelist=0123456789.'
-
-        # Step 6: Run OCR
-        text = pytesseract.image_to_string(thresh, config=config)
-        print("🔍 OCR RAW OUTPUT:", repr(text))  # view this in Hugging Face logs
-
-        # Step 7: Extract numbers
+        # OCR
+        text = pytesseract.image_to_string(blur, config='--psm 7 digits')
         weight = ''.join(filter(lambda c: c in '0123456789.', text))
 
-        confidence = 95 if weight else 0
+        confidence = 95  # Replace with real confidence logic if needed
         return weight.strip(), confidence
-
     except Exception as e:
-        print("❌ OCR Exception:", str(e))
+        print(f"OCR error: {e}")
         return "", 0