MMDeploy bindings for Rust.

This repo is a low-level MMDeploy>=1.0.0 abstraction, you can use rust-mmdeploy directly if you want to use MMDeploy in Rust. If it cannot cover your requirements, it's easy to build high-level abstraction based on this repo.

Supporting all devices and platforms is a big challenge for this personal project. So I just support and test for main devices and platforms.

Supported : ✔️ MMDeploy not support : - Uncertain (without test): ❓

In order to successfully build this repo, you are supposed to install some pre-packages.

Step 1. Install Clang required by Bindgen.

apt install llvm-dev libclang-dev clang

Step 2(Pre-built package). Download and install pre-built mmdeploy package and ONNXRuntime.

wget https://github.com/open-mmlab/mmdeploy/releases/download/v1.1.0/mmdeploy-1.1.0-linux-x86_64-cuda11.3.tar.gz
tar -zxvf mmdeploy-1.1.0-linux-x86_64-cuda11.3.tar.gz
cd mmdeploy-1.1.0-linux-x86_64-cuda11.3.tar.gz
export MMDEPLOY_DIR=$(pwd)
export ONNXRUNTIME_DIR=$(pwd)/thirdparty/onnxruntime
export LD_LIBRARY_PATH=$ONNXRUNTIME_DIR/lib:$LD_LIBRARY_PATH
export TENSORRT_DIR=$(pwd)/thirdparty/tensorrt
export LD_LIBRARY_PATH=$TENSORRT_DIR/lib:$LD_LIBRARY_PATH

Then follow the guide in $MMDEPLOY_DIR/README.md to build the SDK. **Step 2(Build from source).**Follow the official guide to build MMDeploy SDK from source. If successfully built, you should have set TENSORRT_DIR, ONNXRUNTIME_DIR and LD_LIBRARY_PATH environment variables. Then

export MMDEPLOY_DIR=/path/to/mmdeploy/build/install
export LD_LIBRARY_PATH=$MMDEPLOY_DIR/lib:$LD_LIBRARY_PATH

Update your Cargo.toml

mmdeploy-sys = "1.1.0"

Move to the root of this repo and then:

cargo build