Project Introduction

This project is based on CNN+BLSTM+CTC to realize verification code identification. This project is only for deployment models, If you need to train the model, please move to https://github.com/kerlomz/captcha_trainer

Informed

1. The default requirements.txt will install CPU version, Change "requirements.txt" from "TensorFlow" to "TensorFlow-GPU" to Switch to GPU version, Use the GPU version to install the corresponding CUDA and cuDNN.
2. demo.py: An example of how to call a prediction method.
3. The model folder folder is used to store model configuration files such as model.yaml.
4. The graph folder is used to store compiled models such as model.pb
5. The deployment service will automatically load all the models in the model configuration. When a new model configuration is added, the corresponding compilation model in the graph folder will be automatically loaded, so if you need to add it, please copy the corresponding compilation model to the graph path first, then add the model configuration.

Start

1. Install the python 3.9 environment (with pip)
2. Install virtualenv pip3 install virtualenv
3. Create a separate virtual environment for the project:

   virtualenv -p /usr/bin/python3 venv # venv is the name of the virtual environment.
   cd venv/ # venv is the name of the virtual environment.
   source bin/activate # to activate the current virtual environment.
   cd captcha_platform # captcha_platform is the project path.

4. pip install -r requirements.txt
5. Place your trained model.yaml in model folder, and your model.pb in graph folder (create if not exist)
6. Deploy as follows.

1. Http Version

1. Linux Deploy (Linux/Mac):

   Port: 19952

   python tornado_server.py
   

2. Windows Deploy (Windows):

   python xxx_server.py
   

3. Request

   Request URI	Content-Type	Payload Type	Method
   http://localhost:[Bind-port]/captcha/v1	application/json	JSON	POST

   Parameter	Required	Type	Description
   image	Yes	String	Base64 encoding binary stream
   model_name	No	String	ModelName, bindable in yaml configuration

   The request is in JSON format, like: {"image": "base64 encoded image binary stream"}

4. Response

   Parameter Name	Type	Description
   message	String	Identify results or error messages
   code	String	Status Code
   success	String	Whether to request success

   The return is in JSON format, like: {"message": "xxxx", "code": 0, "success": true}

2. G-RPC Version

Deploy:

python3 grpc_server.py

Port: 50054

Update G-RPC-CODE

python -m grpc_tools.protoc -I. --python_out=. --grpc_python_out=. ./grpc.proto

Directory Structure

- captcha_platform
    - grpc_server.py
    - flask_server.py
    - tornado_server.py
    - sanic_server.py
    - demo.py
    - config.yaml
- model
    - model-1.yaml
    - model-2.yaml
    - ...
- graph
    - Model-1.pb
    - ...

Management Model

1. Load a model

• Put the trained pb model in the graph folder.
• Put the trained yaml model configuration file in the model folder.

2. Unload a model

• Delete the corresponding yaml configuration file in the model folder.
• Delete the corresponding pb model file in the graph folder.

3. Update a model

• Put the trained pb model in the graph folder.
• Put the yaml configuration file with "Version" greater than the current version in the model folder.
• Delete old models and configurations.

License

This project use SATA License (Star And Thank Author License), so you have to star this project before using. Read the license carefully.

Introduction

https://www.jianshu.com/p/80ef04b16efc

Donate

Thank you very much for your support of my project.