Bump num_inference from 10 to 2000

Change-Id: Iad29b917c9d7e9c286f675cf1dc4a2bc67f0cf5a
(cherry picked from commit d2a8fa9eaaafc995325d72afcbb605301fc5d1ef)
1 file changed
tree: 9b63e7ea2016be25e18f3d58e936b65b4b3b419b
  1. .gitignore
  2. 99-edgetpu-accelerator.rules
  5. MANIFEST.in
  6. README.md
  7. benchmarks/
  8. build_package.sh
  9. compiler/
  10. debian/
  11. docs/
  12. edgetpu/
  13. install.sh
  14. libedgetpu/
  15. qa_test.sh
  16. setup.py
  17. test_data/
  18. tests/
  19. uninstall.sh

Edge TPU Python API

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

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. cd packages/edgetpu/

  3. ./build_package.sh

  4. tar xzf edgetpu_api_<version>.tar.gz

  5. cd edgetpu_api/

  6. sudo ./install.sh

  7. Now check for the installed library at /usr/local/lib/python3.6/dist-packages/edgetpu/

If it seems the library did not update compared to an older version you had installed, then run the uninstall.sh script and then rerun install.sh (it probably skipped installing the new library because the version number hasn't been updated yet).