Installation Instructions

Precompiled libraries in package managers(Nuget, PYPI, etc)

Core part of BrainFlow is written in C/C++ and distributed as dynamic libraries, for some programming languages we publish packages with precompiled libraries to package managers like Nuget or PYPI.

C/C++ code should be compiled for each CPU architecture and for each OS and we cannot cover all possible cases, as of right now we support:

  • x64 libraries for Windows starting from 8.1, for some devices newer version of Windows can be required

  • x64 libraries for Linux, they are compiled inside manylinux docker container

  • x64/ARM libraries for MacOS, they are universal binaries

If your CPU and OS is not listed above(e.g. Raspberry Pi or Windows with ARM) you still can use BrainFlow, but you need to compile it by youself first. See Compilation of Core Module and C++ Binding for details.

Python

Please, make sure to use Python 3+. Next, install the latest release from PYPI with the following command in terminal

python -m pip install brainflow

If you want to install it from source files or build unreleased version from Github, you should first compile the core module (Compilation of Core Module and C++ Binding). Then run

cd python_package
python -m pip install -U .

C#

Windows(Visual Studio)

You are able to install the latest release from Nuget or build it yourself:

  • Compile BrainFlow’s core module

  • Open Visual Studio Solution

  • Build it using Visual Studio

  • Make sure that unmanaged(C++) libraries exist in search path - set PATH env variable or copy them to correct folder

Unix(Mono)

  • Compile BrainFlow’s core module

  • Install Mono and other dependcies on your system

  • Build it using dotnet command

  • Make sure that unmanaged(C++) libraries exist in search path - set LD_LIBRARY_PATH env variable or copy them to correct folder

Example for Fedora:

# compile c++ code
python tools/build.py
# install dependencies
sudo dnf install nuget
sudo dnf install mono-devel
sudo dnf install mono-complete
sudo dnf install monodevelop
sudo dnf install dotnet-sdk-6.0
sudo dnf install dotnet-runtime-6.0
sudo dnf install dotnet-sdk-3.1
sudo dnf install dotnet-runtime-3.1
# build solution
dotnet build csharp_package/brainflow/brainflow.sln
# run tests
export LD_LIBRARY_PATH=/home/andreyparfenov/brainflow/installed/lib/
mono csharp_package/brainflow/examples/denoising/bin/Debug/denoising.exe

R

R binding is based on reticulate package and calls Python , so you need to install Python binding first, make sure that reticulate uses correct virtual environment, after that you will be able to build R package from command line or using R Studio, install it and run samples.

Java

You are able to download jar files directly from release page

If you want to install it from source files or build unreleased version from github you should compile core module first (Compilation of Core Module and C++ Binding) and run

cd java_package
cd brainflow
mvn package

Also, you can use GitHub Package and download BrainFlow using Maven or Gradle. To use Github packages you need to change Maven settings. Example file here you need to change OWNER and TOKEN by Github username and token with an access to Github Packages.

Matlab

Steps to setup Matlab binding for BrainFlow:

  • Compile Core Module, using the instructions in Compilation of Core Module and C++ Binding. If you don’t want to compile C++ code you can download Matlab package with precompiled libs from Release page

  • Open Matlab IDE and open brainflow/matlab_package/brainflow folder there

  • Add folders lib and inc to Matlab path

  • If you want to run Matlab scripts from folders different than brainflow/matlab_package/brainflow you need to add it to your Matlab path too

  • If you see errors you may need to configure Matlab to use C++ compiler instead C, install Visual Studio 2017 or newer(for Windows) and run this command in Matlab terminal mex -setup cpp, you need to select Visual Studio Compiler from the list. More info can be found here.

Julia

BrainFlow is a registered package in the Julia general registry, so it can be installed via the Pkg manager:

Example:

import Pkg
Pkg.add("BrainFlow")

When using BrainFlow for the first time in Julia, the BrainFlow artifact containing the compiled BrainFlow libraries will be downloaded from release page automatically.

If you compile BrainFlow from source local libraries will take precedence over the artifact.

Typescript

You can install BrainFlow using next command without compilation

npm install brainflow

If you want to install it from source files or build unreleased version from Github, you should first compile the core module (Compilation of Core Module and C++ Binding). Then run

cd nodejs_package
npm install

Rust

You can build Rust binding locally using commands below, but you need to compile C/C++ code first

cd rust_package
cd brainflow
cargo build --features generate_binding

Docker Image

There are docker images with precompiled BrainFlow. You can get them from DockerHub.

All bindings except Matlab are preinstalled there.

Also, there are other packages for BCI research and development:

  • mne

  • pyriemann

  • scipy

  • matplotlib

  • jupyter

  • pandas

  • etc

If your devices uses TCP/IP to send data, you need to run docker container with --network host. For serial port connection you need to pass serial port to docker using --device %your port here%

Example:

# pull container from DockerHub
docker pull brainflow/brainflow:latest
# run docker container with serial port /dev/ttyUSB0
docker run -it --device /dev/ttyUSB0 brainflow/brainflow:latest /bin/bash
# run docker container for boards which use networking
docker run -it --network host brainflow/brainflow:latest /bin/bash

Compilation of Core Module and C++ Binding

Windows

  • Install CMake>=3.16 you can install it from PYPI via pip or from CMake website

  • Install Visual Studio 2019(preferred) or Visual Studio 2017. Other versions may work but not tested

  • In VS installer make sure you selected “Visual C++ ATL support”

  • Build it as a standard CMake project, you don’t need to set any options

If you are not familiar with CMake you can use build.py :

# install python3 and run
python -m pip install cmake
cd tools
python build.py
# to get info about args and configure your build you can run
python build.py --help

Linux

  • Install CMake>=3.16 you can install it from PYPI via pip, via package managers for your OS(apt, dnf, etc) or from CMake website

  • If you are going to distribute compiled Linux libraries you HAVE to build it inside manylinux Docker container

  • Build it as a standard CMake project, you don’t need to set any options

  • You can use any compiler but for Linux we test only GCC

If you are not familiar with CMake you can use build.py :

python3 -m pip install cmake
cd tools
python3 build.py
# to get info about args and configure your build you can run
python3 build.py --help

MacOS

  • Install CMake>=3.16 you can install it from PYPI via pip, using brew or from CMake website

  • Build it as a standard CMake project, you don’t need to set any options

  • You can use any compiler but for MacOS we test only Clang

If you are not familiar with CMake you can use build.py :

python3 -m pip install cmake
cd tools
python3 build.py
# to get info about args and configure your build you can run
python3 build.py --help

Android

To check supported boards for Android visit Supported Boards

Installation instructions

  • Create Java project in Android Studio, Kotlin is not supported

  • Download jniLibs.zip from Release page

  • Unpack jniLibs.zip and copy it’s content to project/app/src/main/jniLibs

  • Download brainflow-jar-with-dependencies.jar from Release page or from Github package

  • Copy brainflow-jar-with-dependencies.jar to project/app/libs folder

Now you can use BrainFlow SDK in your Android application!

Note: Android Studio inline compiler may show red errors but it should be compiled fine with Gradle. To fix inline compiler you can use File > Sync Project with Gradle Files or click at File > Invalidate Cache/Restart > Invalidate and Restart

Prebuild libraries for jniLibs.zip are complied using:

  • Android NDK 25.1.8937393

  • -DANDROID_NATIVE_API_LEVEL=android-24

For some API calls you need to provide additional permissions via manifest file of your application

<uses-permission android:name="android.permission.INTERNET"></uses-permission>
<uses-permission android:name="android.permission.ACCESS_NETWORK_STATE"></uses-permission>
<uses-permission android:name="android.permission.READ_EXTERNAL_STORAGE"></uses-permission>
<uses-permission android:name="android.permission.WRITE_EXTERNAL_STORAGE"></uses-permission>

Compilation using Android NDK

For BrainFlow developers

To test your changes in BrainFlow on Android you need to build it using Android NDK manually.

Compilation instructions:

  • Download Android NDK

  • Download Ninja or get one from the tools folder, make sure that ninja.exe is in search path

  • You can also try MinGW Makefiles instead Ninja, but it’s not tested and may not work

  • Build C++ code using cmake and Ninja for all ABIs

  • Compiled libraries will be in tools/jniLibs folder

Command line examples:

# to prepare project(choose ABIs which you need)
# for arm64-v8a
cmake -G Ninja -DCMAKE_TOOLCHAIN_FILE=D:\workspace\android-ndk-r25b\build\cmake\android.toolchain.cmake -DANDROID_NATIVE_API_LEVEL=android-24 -DANDROID_ABI=arm64-v8a ..
# for armeabi-v7a
cmake -G Ninja -DCMAKE_TOOLCHAIN_FILE=D:\workspace\android-ndk-r25b\build\cmake\android.toolchain.cmake -DANDROID_NATIVE_API_LEVEL=android-24 -DANDROID_ABI=armeabi-v7a ..
# for x86_64
cmake -G Ninja -DCMAKE_TOOLCHAIN_FILE=D:\workspace\android-ndk-r25b\build\cmake\android.toolchain.cmake -DANDROID_NATIVE_API_LEVEL=android-24 -DANDROID_ABI=x86_64 ..
# for x86
cmake -G Ninja -DCMAKE_TOOLCHAIN_FILE=D:\workspace\android-ndk-r25b\build\cmake\android.toolchain.cmake -DANDROID_NATIVE_API_LEVEL=android-24 -DANDROID_ABI=x86 ..

# to build(should be run for each ABI from previous step**
cmake --build . --target install --config Release -j 2 --parallel 2