Check the preview of 2nd version of this platform being developed by the open MLCommons taskforce on automation and reproducibility as a free, open-source and technology-agnostic on-prem platform.

Object detection • MLPerf inference • TensorFlow CPU • COCO • 50 images validation • MacOS • benchmark • Portable Workflows

solution:demo-obj-detection-coco-tf-cpu-benchmark-macos-portable-workflows (v1.0.1)

Portable solution description  

Install and run this solution on your platform in several simple steps. Our goal is to make it simpler to reproduce results from research papers, participate in crowd-benchmarking, and enable "live" papers.
Don't hesitate to get in touch if you encounter any issues or would like to discuss this community project!

Check the prerequisites for your system  

Install manually from the command line (to be automated in the future):
# Tested on Apple MacBook Pro 11.1

# These dependencies are needed to rebuild COCO API:

 brew install git wget zlib curl cmake
 brew install gcc autoconf autogen libtool
 brew install freetype
 brew install llvm

# You also likely need to install Xcode from Apple app store

# You may also need to install wget with SSL
 brew uninstall wget
 brew install wget --with-libressl

Install cBench (docs)

Install cBench from the command line (a small Python library to manage CK solutions):
pip3 install cbench
 or 
python3 -m pip install cbench
 or
pip install cbench 
Note that you may need to add the --user flag if you install in your user space, i.e. "python3 -m pip install cbench --user"

Init this solution with the portable workflow on your machine

Run manually from your command line (cBench will attempt to automatically adapt this workflow to your system - you may need to press Enter several times to select default answers for some questions):
cb init demo-obj-detection-coco-tf-cpu-benchmark-macos-portable-workflows

Start cBench (status: disconnected)

cb start

Participate in collaborative crowd-benchmarking of this solution

or start crowd-benchmarking manually from the command line:
cb benchmark demo-obj-detection-coco-tf-cpu-benchmark-macos-portable-workflows

Run this workflow locally

or start local run manually from the command line:
cb run demo-obj-detection-coco-tf-cpu-benchmark-macos-portable-workflows

  # Note that the following CK program pipeline will be executed:
  ck compile program:object-detection-tf-py-benchmark --cmd_key=default --speed
  ck run program:object-detection-tf-py-benchmark --cmd_key=default

Successfully tested configuration

Host OS: linux-64 (Ubuntu 18.04.3 LTS)
Target OS: linux-64 (Ubuntu 18.04.3 LTS)
Target CPU: Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz
Target CPUs:
Python min version: 3.6
Python max version: 3.7.99

Dependencies    

Reused CK components

These components are automatically installed by cBench from this portal:
cb download --version=1.0.0 --force package:model-tf-mlperf-ssd-mobilenet
cb download --version=1.0.0 --force soft:model.tensorflow.object-detection
cb download --version=1.0.0 --force script:install-package-tensorflowmodel-object-detection

cb download --version=1.0.0 --force package:lib-python-numpy
cb download --version=1.0.0 --force soft:lib.python.numpy

cb download --version=1.0.0 --force package:lib-python-scipy
cb download --version=1.0.0 --force soft:lib.python.scipy

cb download --version=1.0.0 --force package:lib-python-matplotlib
cb download --version=1.0.0 --force soft:lib.python.matplotlib

cb download --version=1.0.0 --force package:lib-python-pillow
cb download --version=1.0.0 --force soft:lib.python.pillow

cb download --version=1.0.0 --force package:lib-python-cython
cb download --version=1.0.0 --force soft:lib.python.cython

cb download --version=1.0.0 --force package:lib-python-cv2
cb download --version=1.0.0 --force soft:lib.python.cv2

cb download --version=1.0.0 --force package:tool-coco-codereef
cb download --version=1.0.0 --force soft:tool.coco

cb download --version=1.0.1 --force package:dataset-coco-2017-val-small
cb download --version=1.0.0 --force soft:dataset.coco.2017.val

cb download --version=1.0.0 --force package:lib-tensorflow-1.1.0-cpu
cb download --version=1.0.0 --force package:lib-tensorflow-1.4.0-cpu
cb download --version=1.0.0 --force package:lib-tensorflow-1.14.0-cpu
cb download --version=1.0.0 --force soft:lib.tensorflow

cb download --version=1.0.0 --force package:labelmap-coco
cb download --version=1.0.0 --force soft:labelmap.object-detection

cb download --version=1.0.0 --force soft:compiler.gcc
cb download --version=1.0.0 --force soft:compiler.llvm

cb download --version=1.0.0 --force package:tensorflowmodel-api-object-detection
cb download --version=1.0.0 --force soft:model.tensorflow-models-api

cb download --version=1.0.0 --force package:lib-protobuf-3.0.0-host
cb download --version=1.0.0 --force soft:lib.protobuf.host

cb download --version=1.0.0 --force script:process-compiler-for-cmake

cb download --version=1.0.0 --force soft:tool.cmake

cb download --version=1.0.0 --force soft:lib.protobuf.host

cb download --version=1.0.1 --force module:program
cb download --version=1.0.0 --force module:dataset
cb download --version=1.0.0 --force module:pipeline
cb download --version=1.0.0 --force module:choice
cb download --version=1.0.0 --force module:experiment
cb download --version=1.0.0 --force module:math.variation

cb download --version=1.2.1 --force program:object-detection-tf-py-benchmark




ck install package:lib-tensorflow-1.14.0-cpu

ck install package --tags=model,tf,object-detection,mlperf,ssd-mobilenet,non-quantized

ck install package --tags=lib,python-package,numpy
ck install package --tags=lib,python-package,scipy --force_version=1.2.1
ck install package --tags=lib,python-package,matplotlib
ck install package --tags=lib,python-package,pillow
ck install package --tags=lib,python-package,cython
ck install package --tags=lib,python-package,cv2

ck install package:tool-coco-codereef

ck install package:dataset-coco-2017-val-small

Comments  

Please log in to add your comments!
If you notice any inapropriate content that should not be here, please report us as soon as possible and we will try to remove it within 48 hours!