feat: update version to 0.2.3 and refactor captcha processing to use OcrModel

Author: thedtvn

README.MD

@@ -1,17 +1,17 @@
 # MBBank API for Python
-A unofficially light weight Python Api for the "Military Commercial Joint Stock Bank" personal account
+A unofficially lightweight Python Api for the "Military Commercial Joint Stock Bank" personal account
 
 ## Why we  have this lib ?
 
 Because I don't want to lose money on some api services like [casso.vn](https://casso.vn/) or [payos.vn](https://payos.vn/)
 when have freemium tier but with limit :>
 
-what you need that it use this lib to make your own code so you can use it free forever
+What you need that it use this lib to make your own code so you can use it free forever
 
 [Documentation & FAQ](http://mbbank.rtfd.io/)
 
 ## Requires
-   - [pytorch with cuda](https://pytorch.org/get-started/locally/#start-locally) (optional if you want to use cuda gpu)
+   - [onnx with cuda](https://onnxruntime.ai/getting-started) (optional if you want to use cuda gpu)
 
    - python3
 
@@ -29,4 +29,6 @@ what you need that it use this lib to make your own code so you can use it free
 
 - [MBBank JS API](https://github.com/CookieGMVN/MBBank)
 
+- [MBBank OCR](https://github.com/thedtvn/mbbank-capcha-ocr)
+
 ![Hits](https://hits.seeyoufarm.com/api/count/incr/badge.svg?url=https%3A%2F%2Fgithub.com%2Fthedtvn%2FMBBank&count_bg=%2379C83D&title_bg=%23555555&icon=python.svg&icon_color=%23E7E7E7&title=Views+Counter&edge_flat=false)

docs/index.md

@@ -9,7 +9,7 @@ when have freemium tier but with limit :>
 what you need that it use this lib to make your own code so you can use it free forever
 
 ## Requires
-   - [pytorch with cuda](https://pytorch.org/get-started/locally/#start-locally) (optional if you want to use cuda gpu)
+   - [onnx with cuda](https://onnxruntime.ai/getting-started) (optional if you want to use cuda gpu)
 
    - python3
 
@@ -35,4 +35,8 @@ These pages go into great detail about everything the API can do.
 
 - Will my account get banned if i use this lib ?
 
-> I can not guarantee that so use it with your own risk.
+> I can not guarantee that so use it with your own risk.
+
+- Can i run ocr model on cpu ?
+
+> Yes, you can. but it maybe will be slower than using gpu (cuda)

mbbank/__init__.py

@@ -3,4 +3,4 @@
 from .wasm_helper import wasm_encrypt
 from .capcha_ocr import CapchaProcessing, CapchaOCR
 
-__version__ = '0.2.2'
+__version__ = '0.2.3'

mbbank/capcha_ocr.py

@@ -1,5 +1,5 @@
 import io
-from mb_capcha_ocr import predict
+from mb_capcha_ocr import OcrModel
 from PIL import Image
 
 
@@ -34,8 +34,16 @@ class CapchaOCR(CapchaProcessing):
     """
     Torch based OCR for capcha processing
     https://pypi.org/project/mb-capcha-ocr/
+
+    Args:
+        model_path (str, optional): path to model file
     """
 
+    def __init__(self, model_path: str = None):
+        super().__init__()
+        # loading model will take about 1-3 seconds
+        self.model = OcrModel(model_path)
+
     def process_image(self, img: bytes) -> str:
         """
         Process image and return text
@@ -47,7 +55,7 @@ def process_image(self, img: bytes) -> str:
             success (str): text from image
         """
         image = Image.open(io.BytesIO(img))
-        text = predict(image)
+        text = self.model.predict(image)
         return text