garyscetbon / edgetpu

Source code for the Edge TPU API library

Edge TPU Python API

This repository contains an easy-to-use Python API to work with Coral devices:

• Dev Board
• USB Accelerator

You can run inference and do transfer learning.

Build and install from source

1. Sync the source code as per the Mendel get started guide.

2. Run cd packages/edgetpu/ to switch current working directory

3. Run scripts/runtime/install.sh to install Edge TPU runtime

4. Run scripts/build_swig.sh to build SWIG-based native layer

5. Run make wheel to generate Python wheel and then pip3 install $(ls dist/*.whl) to install it

Native C++ code

All native code is inside src folder. You can build everything using Makefile. For example, run make tests to build all C++ unit tests or make benchmarks to build all C++ benchmarks. To get the list of all available make targets run make help.

You can cross-compile code for different platforms by setting CPU variable:

make CPU=k8      tests  # Builds for x86_64 (default CPU value)
make CPU=armv7a  tests  # Builds for ARMv7-A, e.g. Pi 3 or Pi 4
make CPU=aarch64 tests  # Builds for ARMv8, e.g. Coral Dev Board

Docker allows to avoid complicated environment setup and run the same Makefile targets:

make DOCKER_IMAGE=debian:buster DOCKER_CPUS="k8 armv7a aarch64" DOCKER_TARGETS=tests docker-build
make DOCKER_IMAGE=ubuntu:18.04  DOCKER_CPUS="k8 armv7a aarch64" DOCKER_TARGETS=tests docker-build

All output goes to out directory.