Spaces:
Sleeping
Sleeping
Fix: Make OCR imports lazy and add version constraints to requirements.txt
Browse files- ocr_strategies.py +34 -4
ocr_strategies.py
CHANGED
|
@@ -1,9 +1,25 @@
|
|
| 1 |
-
import easyocr
|
| 2 |
-
import pytesseract
|
| 3 |
from abc import ABC, abstractmethod
|
| 4 |
from typing import Set
|
| 5 |
-
|
| 6 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
|
| 8 |
class OCRStrategy(ABC):
|
| 9 |
"""Abstract base class for OCR strategies."""
|
|
@@ -14,6 +30,10 @@ class OCRStrategy(ABC):
|
|
| 14 |
class EasyOCRStrategy(OCRStrategy):
|
| 15 |
"""Concrete strategy for EasyOCR."""
|
| 16 |
def __init__(self):
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
# Initialize once to save memory/time
|
| 18 |
print("Loading EasyOCR Model...")
|
| 19 |
self.reader = easyocr.Reader(['en'], gpu=False)
|
|
@@ -24,6 +44,16 @@ class EasyOCRStrategy(OCRStrategy):
|
|
| 24 |
|
| 25 |
class TesseractOCRStrategy(OCRStrategy):
|
| 26 |
"""Concrete strategy for Tesseract OCR (Free & Fast)."""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
def extract_text(self, image_path: str) -> Set[str]:
|
| 28 |
# Preprocessing for better Tesseract accuracy
|
| 29 |
img = cv2.imread(image_path)
|
|
|
|
|
|
|
|
|
|
| 1 |
from abc import ABC, abstractmethod
|
| 2 |
from typing import Set
|
| 3 |
+
|
| 4 |
+
# Lazy imports to handle missing dependencies gracefully
|
| 5 |
+
try:
|
| 6 |
+
import easyocr
|
| 7 |
+
EASYOCR_AVAILABLE = True
|
| 8 |
+
except ImportError:
|
| 9 |
+
EASYOCR_AVAILABLE = False
|
| 10 |
+
|
| 11 |
+
try:
|
| 12 |
+
import pytesseract
|
| 13 |
+
PYTESSERACT_AVAILABLE = True
|
| 14 |
+
except ImportError:
|
| 15 |
+
PYTESSERACT_AVAILABLE = False
|
| 16 |
+
|
| 17 |
+
try:
|
| 18 |
+
import cv2
|
| 19 |
+
import numpy as np
|
| 20 |
+
CV2_AVAILABLE = True
|
| 21 |
+
except ImportError:
|
| 22 |
+
CV2_AVAILABLE = False
|
| 23 |
|
| 24 |
class OCRStrategy(ABC):
|
| 25 |
"""Abstract base class for OCR strategies."""
|
|
|
|
| 30 |
class EasyOCRStrategy(OCRStrategy):
|
| 31 |
"""Concrete strategy for EasyOCR."""
|
| 32 |
def __init__(self):
|
| 33 |
+
if not EASYOCR_AVAILABLE:
|
| 34 |
+
raise ImportError(
|
| 35 |
+
"EasyOCR is not installed. Please install it with: pip install easyocr"
|
| 36 |
+
)
|
| 37 |
# Initialize once to save memory/time
|
| 38 |
print("Loading EasyOCR Model...")
|
| 39 |
self.reader = easyocr.Reader(['en'], gpu=False)
|
|
|
|
| 44 |
|
| 45 |
class TesseractOCRStrategy(OCRStrategy):
|
| 46 |
"""Concrete strategy for Tesseract OCR (Free & Fast)."""
|
| 47 |
+
def __init__(self):
|
| 48 |
+
if not PYTESSERACT_AVAILABLE:
|
| 49 |
+
raise ImportError(
|
| 50 |
+
"pytesseract is not installed. Please install it with: pip install pytesseract"
|
| 51 |
+
)
|
| 52 |
+
if not CV2_AVAILABLE:
|
| 53 |
+
raise ImportError(
|
| 54 |
+
"opencv-python is not installed. Please install it with: pip install opencv-python-headless"
|
| 55 |
+
)
|
| 56 |
+
|
| 57 |
def extract_text(self, image_path: str) -> Set[str]:
|
| 58 |
# Preprocessing for better Tesseract accuracy
|
| 59 |
img = cv2.imread(image_path)
|