Skip to content
Tengine is a lite, high performance, modular inference engine for embedded device
Branch: master
Clone or download
Pull request Compare This branch is 2 commits ahead, 28 commits behind OAID:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.
examples/mobilenet_ssd_yahei init Oct 16, 2018

Tengine for Rasberry Pi(Education Edition)

For education only!

Tengine, developed by OPEN AI LAB, is a lite, high-performance, and modular inference engine for embedded device.


  1. Install tools & libraries

    sudo apt-get install git cmake
    sudo apt-get install libprotobuf-dev protobuf-compiler
    sudo apt-get install libboost-all-dev
    sudo apt-get install libgoogle-glog-dev
    sudo apt-get install libopencv-dev
    sudo apt-get install libopenblas-dev
  2. Get Tengine code

    git clone
  3. Prepare config files

    • copy config example file
      cd ~/Tengine
      cp makefile.config.example makefile.config
    • edit makefile.config
      • comment CONFIG_ARCH_ARM64 option to invalid arm64 arch.
      • uncomment CONFIG_ARCH_BLAS=y option to valid Openblas arch.
  4. Build

    make -j4
  5. Get the binary library
    Replace Tengine/install/lib/ with the new one.


Reference: 树莓派也能玩转深度学习——Tengine推断引擎/测试 | Hey~YaHei!

With Tengine for Rasberry Pi(Education Edition), SPF(Second Per Frame) of mobilenet-ssd model drops from 1.148 to 0.286.

You can’t perform that action at this time.