Computer Vision News - May 2019

Jupyter is a tool for doing interactive computing with human-in-the-loop - something we tend to think of as not very useful and only half-respectable. However, it turns out to be essential for writing some code, inspect the results, and maybe get some rich outputs - in all of our fields: computer science, data science, and even for training networks in computer vision. The Jupyter project started in 2012 with the release of the ipython notebook. JupyterLab is the next generation of this project; it includes all features of the classic Jupyter Notebook with the addition of a flexible, powerful web-based interface which includes: notebook, terminal, text editor, file browser, rich outputs, and much more. It enjoys widespread use, with over 3 million notebooks publicly available on github and a roughly similar number of private notebooks - so this is a tool that has become very popular quickly. Just to give you a quick taste of what it looks like, here is a first print screen from JupyterLab. The first big difference you’ll notice is its multiple window approach, which gives it the look and feel of a “real” IDE. On the left you can see a file browser with a list of files, a list of running notebook sessions and a palette of executable commands. The middle panel should feel familiar to Jupyter ipython notebook users, with a lot of the same features, however, now supporting multiple tabs - multiple notebooks open at the same time. On the right - is the inspector - which you can use as a debugger which lets you inspect the types and values of your variables. On the bottom you can see a bash terminal which lets you run commands from inside the browser. There is also a text editor available. Handy, isn’t it? Installation To set this up on your machine, follow the instructions on the next page: 16 Focus on: JupyterLab Focus on by Assaf Spanier This is Assaf’s last article for Computer Vision News, on which he published almost 100. He now leaves the magazine to join a very different project. We owe to him much of what we achieved and we wish him a lot of success! Computer Vision News