Libraries-and-Extensions-for-TensorFlow-for-Apple-Silicon
Libraries-and-Extensions-for-TensorFlow-for-Apple-Silicon copied to clipboard
Is there a way to run object detection api on Macos 12
Please describe in detail, are you referring to tensorflow-io support?
yes, I am not able to install tensorflow io in my conda environment

❯ python3 -m pip install --no-deps dist/tensorflow_io-0.26.0-cp39-cp39-macosx_11_0_arm64.whl ERROR: tensorflow_io-0.26.0-cp39-cp39-macosx_11_0_arm64.whl is not a supported wheel on this platform.
This approach works with the Object Detection API, but the support is not perfect. For this reason, I haven't written a tensorflow-io tutorial.
-
Use
virtualenvto create an environment (please do not usecondato create new environments!).python3 -m venv ~/tensorflow-macos source ~/tensorflow-macos/bin/activate -
Install
hdf5(please do not usecondapre-compiledhdf5!).brew install hdf5 export CPATH="/opt/homebrew/include/" export HDF5_DIR="/opt/homebrew/" -
Install the
tensorflow-macosandtensorflow-metalplugins.pip install tensorflow-macos==2.9.0 pip install tensorflow-metal==0.5.0 -
Download and extract
io 0.26.0.wget https://github.com/tensorflow/io/archive/refs/tags/v0.26.0.zip unzip v0.26.0.zip cd io-0.26.0 -
Build
io.python setup.py -q bdist_wheel --project tensorflow_io_gcs_filesystem python setup.py -q bdist_wheel -
Install the
whlfile (please install thetensorflow_io_gcs_filesystemfirst, and use the--no-depsparameter).pip install --no-deps ./dist/tensorflow_io_gcs_filesystem-0.25.0-cp39-cp39-macosx_12_0_arm64.whl pip install --no-deps ./dist/tensorflow_io-0.25.0-cp39-cp39-macosx_12_0_arm64.whl
It worked (with a warning) Thank you so much.
I won't close this issue since io can not be built perfectly yet. If you still need it, you can continue to follow it, and I will close it when I can build it perfectly.