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
- 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.
- demo.py: An example of how to call a prediction method.
- The model folder folder is used to store model configuration files such as model.yaml.
- The graph folder is used to store compiled models such as model.pb
- 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.
- Install the python 3.9 environment (with pip)
- Install virtualenv
pip3 install virtualenv
- 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.
pip install -r requirements.txt
- Place your trained model.yaml in model folder, and your model.pb in graph folder (create if not exist)
- Deploy as follows.
-
Linux Deploy (Linux/Mac):
Port: 19952
python tornado_server.py
-
Windows Deploy (Windows):
python xxx_server.py
-
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"}
-
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}
Deploy:
python3 grpc_server.py
Port: 50054
python -m grpc_tools.protoc -I. --python_out=. --grpc_python_out=. ./grpc.proto
- 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
- ...
- Load a model
- Put the trained pb model in the graph folder.
- Put the trained yaml model configuration file in the model folder.
- Unload a model
- Delete the corresponding yaml configuration file in the model folder.
- Delete the corresponding pb model file in the graph folder.
- 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.
This project use SATA License (Star And Thank Author License), so you have to star this project before using. Read the license carefully.
https://www.jianshu.com/p/80ef04b16efc
Thank you very much for your support of my project.