I am running a Machine Learning project using Google Cloud Platform with Tensorflow and Keras. I have about 30,000 PNG images in my dataset. When I run it locally, Keras has great utilities to load images but Google Cloud Services needs to use certain librairies such as tensorflow.file_io (see: Load numpy array in google-cloud-ml job) in order to read in files from a GC bucket.
What is the best way to load the images from a Google Cloud Storage bucket? Right now I am saving them as bytes and reading them in from one file, but it would be great to be able to load the images directly from the GC bucket.
Thanks,
