laugh12321 / model_crypto

A Python and C++ library for model encryption and decryption, built on Crypto++, with support for various deep learning frameworks including TensorRT, PyTorch, Paddle, and more. Protect your AI models with robust encryption techniques.

English | 简体中文

Model Crypto: Python and C++ Library for Model Encryption

Model Crypto is a versatile Python and C++ library for encrypting and decrypting machine learning models. It's built on Crypto++ and offers support for various deep learning frameworks such as TensorRT, PyTorch, Paddle, and more. Use Model Crypto to safeguard your AI models with robust encryption techniques.

Getting Started with XMake

Requirements

• Windows: Microsoft Visual Studio (Tested with Visual Studio 2019, 2022)
• XMake (Latest version recommended)

C++ Compilation

Compile your C++ code with the following commands, specifying the target platform and architecture:

xmake f -p {platform} -a {architecture} -m release
xmake -w

After compilation, you will find the C++ library in the project's root directory, including the include and lib directories. Additionally, the Python library required for building is located in .pyd format within the python\model_crypto\libs directory.

Python Installation

To install the Python library, use the following command:

pip install .

Exmaples

• 模型加密杂谈：TensorRT加密

©️ License

Model Crypto is provided under the MIT.