Spaces:
Sleeping
Sleeping
Update ocr_engine.py
Browse files- ocr_engine.py +12 -6
ocr_engine.py
CHANGED
|
@@ -3,22 +3,27 @@ import numpy as np
|
|
| 3 |
import cv2
|
| 4 |
import re
|
| 5 |
|
| 6 |
-
# Load OCR model once
|
| 7 |
reader = easyocr.Reader(['en'], gpu=False)
|
| 8 |
|
| 9 |
def enhance_image(img):
|
| 10 |
-
#
|
| 11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
|
| 13 |
|
| 14 |
# Denoise
|
| 15 |
gray = cv2.fastNlMeansDenoising(gray, h=15)
|
| 16 |
|
| 17 |
# Sharpen
|
| 18 |
-
kernel = np.array([[0, -1, 0], [-1, 5
|
| 19 |
sharp = cv2.filter2D(gray, -1, kernel)
|
| 20 |
|
| 21 |
-
# Contrast
|
| 22 |
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
|
| 23 |
contrast = clahe.apply(sharp)
|
| 24 |
|
|
@@ -36,7 +41,8 @@ def extract_weight_from_image(pil_img):
|
|
| 36 |
weight_candidates = []
|
| 37 |
|
| 38 |
for _, text, conf in results:
|
| 39 |
-
cleaned = text.lower()
|
|
|
|
| 40 |
cleaned = cleaned.replace("o", "0").replace("s", "5").replace("g", "9")
|
| 41 |
cleaned = re.sub(r"[^\d\.]", "", cleaned)
|
| 42 |
|
|
|
|
| 3 |
import cv2
|
| 4 |
import re
|
| 5 |
|
|
|
|
| 6 |
reader = easyocr.Reader(['en'], gpu=False)
|
| 7 |
|
| 8 |
def enhance_image(img):
|
| 9 |
+
# Resize big images
|
| 10 |
+
max_dim = 1000
|
| 11 |
+
height, width = img.shape[:2]
|
| 12 |
+
if max(height, width) > max_dim:
|
| 13 |
+
scale = max_dim / max(height, width)
|
| 14 |
+
img = cv2.resize(img, None, fx=scale, fy=scale, interpolation=cv2.INTER_AREA)
|
| 15 |
+
|
| 16 |
+
# Grayscale
|
| 17 |
gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
|
| 18 |
|
| 19 |
# Denoise
|
| 20 |
gray = cv2.fastNlMeansDenoising(gray, h=15)
|
| 21 |
|
| 22 |
# Sharpen
|
| 23 |
+
kernel = np.array([[0, -1, 0], [-1, 5,-1], [0, -1, 0]])
|
| 24 |
sharp = cv2.filter2D(gray, -1, kernel)
|
| 25 |
|
| 26 |
+
# Contrast
|
| 27 |
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
|
| 28 |
contrast = clahe.apply(sharp)
|
| 29 |
|
|
|
|
| 41 |
weight_candidates = []
|
| 42 |
|
| 43 |
for _, text, conf in results:
|
| 44 |
+
cleaned = text.lower()
|
| 45 |
+
cleaned = cleaned.replace("kg", "").replace("kgs", "")
|
| 46 |
cleaned = cleaned.replace("o", "0").replace("s", "5").replace("g", "9")
|
| 47 |
cleaned = re.sub(r"[^\d\.]", "", cleaned)
|
| 48 |
|