Create ocr_engine.py

ocr_engine.py

@@ -0,0 +1,55 @@
+from mmocr.utils.ocr import MMOCR
+import numpy as np
+import cv2
+import re
+from PIL import Image
+
+ocr = MMOCR(det='DBPANet', recog='SAR', device='cpu')
+
+def extract_weight_from_image(pil_img):
+    try:
+        img = np.array(pil_img.convert("RGB"))[:, :, ::-1]
+        result = ocr.readtext(img, print_result=False, output=None)[0]['result']
+
+        raw_texts = []
+        weight_candidates = []
+        fallback_weight = None
+        fallback_conf = 0.0
+
+        for text, conf in result:
+            original = text
+            cleaned = text.lower().strip()
+            cleaned = cleaned.replace(",", ".")
+            cleaned = cleaned.replace("o", "0").replace("O", "0")
+            cleaned = cleaned.replace("s", "5").replace("S", "5")
+            cleaned = cleaned.replace("g", "9").replace("G", "6")
+            cleaned = cleaned.replace("kg", "").replace("kgs", "")
+            cleaned = re.sub(r"[^0-9\.]", "", cleaned)
+
+            raw_texts.append(f"{original} → {cleaned} (conf: {round(conf, 2)})")
+
+            if cleaned and cleaned.replace(".", "").isdigit() and not fallback_weight:
+                fallback_weight = cleaned
+                fallback_conf = conf
+
+            if cleaned.count(".") <= 1 and re.fullmatch(r"\d{2,4}(\.\d{1,3})?", cleaned):
+                weight_candidates.append((cleaned, conf))
+
+        if weight_candidates:
+            best_weight, best_conf = sorted(weight_candidates, key=lambda x: -x[1])[0]
+        elif fallback_weight:
+            best_weight, best_conf = fallback_weight, fallback_conf
+        else:
+            return "Not detected", 0.0, "\n".join(raw_texts)
+
+        if "." in best_weight:
+            int_part, dec_part = best_weight.split(".")
+            int_part = int_part.lstrip("0") or "0"
+            best_weight = f"{int_part}.{dec_part}"
+        else:
+            best_weight = best_weight.lstrip("0") or "0"
+
+        return best_weight, round(best_conf * 100, 2), "\n".join(raw_texts)
+
+    except Exception as e:
+        return f"Error: {str(e)}", 0.0, "OCR failed"