Update ocr_engine.py

ocr_engine.py

@@ -3,28 +3,21 @@ import numpy as np
 import cv2
 import re
 
-# Initialize EasyOCR
 reader = easyocr.Reader(['en'], gpu=False)
 
 def enhance_image(img):
-    # Downscale large images
     max_dim = 1000
     height, width = img.shape[:2]
     if max(height, width) > max_dim:
         scale = max_dim / max(height, width)
         img = cv2.resize(img, None, fx=scale, fy=scale, interpolation=cv2.INTER_AREA)
 
-    # Convert to grayscale
     gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
-
-    # Denoise
     gray = cv2.fastNlMeansDenoising(gray, h=15)
 
-    # Sharpen
     kernel = np.array([[0, -1, 0], [-1, 5, -1], [0, -1, 0]])
     sharp = cv2.filter2D(gray, -1, kernel)
 
-    # Enhance contrast
     clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
     enhanced = clahe.apply(sharp)
 
@@ -44,7 +37,7 @@ def extract_weight_from_image(pil_img):
         for _, text, conf in results:
             cleaned = text.lower().strip()
 
-            # Replace common OCR misreads
+            # Fix common OCR errors
             cleaned = cleaned.replace(",", ".")
             cleaned = cleaned.replace("o", "0").replace("O", "0")
             cleaned = cleaned.replace("s", "5").replace("S", "5")
@@ -52,18 +45,23 @@ def extract_weight_from_image(pil_img):
             cleaned = cleaned.replace("kg", "").replace("kgs", "")
             cleaned = re.sub(r"[^\d\.]", "", cleaned)
 
-            # Match numbers like 84.5, 102.3, 99.9
-            if re.fullmatch(r"\d{2,4}(\.\d{1,2})?", cleaned):
+            # Match weights like: 58.8, 75.02, 97.2, 102.34, etc.
+            if re.fullmatch(r"\d{2,4}(\.\d{1,3})?", cleaned):
                 weight_candidates.append((cleaned, conf))
 
         if not weight_candidates:
             return "Not detected", 0.0, "\n".join(ocr_texts)
 
-        # Pick highest confidence match
+        # Pick the highest confidence result
         best_weight, best_conf = sorted(weight_candidates, key=lambda x: -x[1])[0]
 
-        # Strip leading zeros if any
-        best_weight = best_weight.lstrip('0') or '0'
+        # Remove unnecessary leading zeros
+        if "." in best_weight:
+            parts = best_weight.split(".")
+            parts[0] = parts[0].lstrip("0") or "0"
+            best_weight = ".".join(parts)
+        else:
+            best_weight = best_weight.lstrip("0") or "0"
 
         return best_weight, round(best_conf * 100, 2), "\n".join(ocr_texts)