Posted on

Proposed New Workflow

With the new Super Resolution Neural Net underway it may be worth looking at a new workflow for the lab because the new neural net is hosted on the Google Colab platform. Below is the diagram for the new workflow that takes advantage of the fact that the new NN is cloud-based.

Fig 1: New Workflow

Workflow Explained

Step 1: Run Data Collection

Step 2: Upload data to google drive

Step 3: Run NN which will pull from data stored on google drive

Step 4: Take the output of the NN and use it to do single image autofocusing over a single region

Step 5: Send that output to the LED array controller to facilitate the autofocusing

Step 6: Run more data collection

Posted on

Should You Switch To TensorFlow 2.0

TensorFlow 2.0 officially released on September 30th, 2019. If you’re about to start a new ML project like me you may be a bit confused about which version of TensorFlow you should use. After all, Python 3 was released in 2008 and support for Python 2.7 will finally case on January 1st, 2020. Granted, that’s not a perfect comparison but TensorFlow 1.x has been around since 2015 and there are numerous articles, tutorials, and support forums to guide you to success using TensorFlow 1.x.  I’m here to tell you that you should switch to TensorFlow 2.0 for your next machine learning project.

A Few Of The Key Changes In TensorFlow 2.0

In short, TensorFlow 2.0 is more user-friendly and easier to debug than TensorFlow 1.x. This all comes down to the fact that TensorFlow 2.0 has eager execution by default. This means that you no longer have to stitch together the graph by making tf.*! In addition, there are no more globals meaning it is now easier to keep track of your variables and their scopes.  Last, but certainly not least is that TensorFlow 2.0 will support legacy code from TensorFlow 1.x.

With all that out of the way, it’s easy to see that TensorFlow 2.0 is a good choice for your next project. Now on to learning TensorFlow 2.0.

Tutorials And DocumentationExist For TensorFlow 2.0!

freeCodeCamp’s Tutorial

Coursera Course

MNIST W/ Google Collab

Documentation

Sources And More Reading

https://towardsdatascience.com/whats-new-in-tensorflow-2-0-ce75cdd1a4d1

https://www.tensorflow.org/guide/effective_tf2

https://medium.com/tensorflow/effective-tensorflow-2-0-best-practices-and-whats-changed-a0ca48767aff

https://hackernoon.com/everything-you-need-to-know-about-tensorflow-2-0-b0856960c074