Welcome. I am an experimental physicist interested in experimental design, data, science and other things. My current goal is to explore fun hardware projects and data science, in particular machine learning.
A hummingbird feeder instrumented with a Raspberry Pi captures images of hummingbirds. These images are processed by a machine learning pipeline. The first step is to classify each image according to if there is indeed a bird in the image or not. For this, a training set of about 3800 images were labeled and are being used to test multiple classifiers.
The code is on Github: https://github.com/Freija/birdometer-ml.
Check out all details on the project page.
I built a website to track my father’s progress as he is biking through the Andes from Ecuador to the very bottom of Argentina. He has a satellite phone that he uses to send a text message with his GPS coordinates, typically once a day. The website is private and used by family members to follow the adventure.
Whereispatrick.at displays the GPS coordinates on a Google Map. It also shows the locations of any geo-tagged photos that my father uploads to Google Drive. Thumbnails of the photos are shown when clicking on the photo markers.
This is a Flask website and API. The text messages are handled by Twilio.The sklearn DBSCAN clustering algorithm is used to cluster photos that are taken close together. Docker is used for testing and deployment.
The code is on Github: https://github.com/Freija/whereispatrick.
Check out all details on the project page.