In the last post, Testing Google Vision API, the different ways to access the Google Vision API to label photos were discussed. As examples, some of the hummingbird images from the birdometer project were fed to the API. In this post, we will loop over all the images in the birdometer training set and collect the labels for each one.
For the birdometer project, one of the critical steps is to identify whether or not a picture has a hummingbird in it or not. There are multiple Vision APIs available that we could potentially use for this task. This blog post discusses the tests done with the Google Vision API.
As part of a data-science workshop in August 2016 organized at UC Berkeley, my team and I wrote a recommendation engine that is accessible through a Flask API. The app runs inside a Docker container and was initially deployed on an AWS EC2 instance. This post discusses deploying the code on Heroku using Docker.
When I was working at the Lawrence Berkeley National Lab, I started the wall of fame/shame called the Ugly Plot Spot.