Description

Project is supposed to be a high-performance backend, which will process videofiles, pictures or interact with user via frontend part. Frontend part is here - https://github.com/valensys/ai-inference-frontend

Prepare Ubuntu (tested on 22.04 and 24.04)

Install packages

sudo apt install libeigen3-dev
sudo apt install nlohmann-json3-dev

Move to directory /usr/local/src and clone dependencies

git clone https://github.com/nadjieb/cpp-mjpeg-streamer.git
git clone -b 4.7.0 https://github.com/opencv/opencv.git
git clone -b poco-1.13.3-release https://github.com/pocoproject/poco.git

Also download folder with pre-built libraries

wget https://github.com/microsoft/onnxruntime/releases/download/v1.17.3/onnxruntime-linux-x64-1.17.3.tgz

Build and install libraries inside /usr/local/src directory

MJPG streamer

cd cpp-mjpeg-streamer && mkdir build && cd build
cmake ..
make && make install
cd ../../

POCO

cd poco && mkdir cmake_build && cd cmake_build
cmake ..
make && make install
cd ../../

OpenCV

cd opencv && mkdir build && cd build
cmake -DCMAKE_CXX_STANDARD=17 ..
make && make install
cd ../../

Extract the archive with OnnxRuntime and move it to /usr/local/onnxruntime

tar xfv onnxruntime-linux-x64-1.17.3.tgz -C /usr/local
ln -s /usr/local/onnxruntime-linux-x64-1.17.3 /usr/local/onnxruntime

Move to /opt directory and clone a branch of "recognition" repository there

cd /opt
git clone -b master --single-branch --depth=1 https://github.com/valensys/ai-inference-backend.git

Build backend

cd /opt/ai-inference-backend
mkdir build && cd build
cmake ..
make

Set up system variable LD_LIBRARY_PATH (if it's not set yet)

export LD_LIBRARY_PATH=/usr/local/lib

Change config file and launch binary

Check recognition/configs/main.cfg and change "host" and "port" according to requirements. For example assign: host=10.128.0.9, port=5200

./Backend