Welcome to the 4IR CDT Santa Hack 2018!
The CDT Christmas hackathon is designed to help you improve your data science skills in a friendly and supportive environment. At the hackathon, you can either work on your own or in a team of up to three people. The hack challenge is to classify the test images as Santa or Not Santa and the team with the highest accuracy at the end of the hack wins. Simples.
The hack agenda and instructions for what you might need to bring are here.
There is no single way to win the Santa Hack - you can use whichever machine learning approach you want: SVM, Random Forests, CNNs... and you can extract whatever features you like along the way. The only thing you do have to do is produce a piece of code that classifies the approved Santa Hack test image dataset and (honestly!) reports its accuracy. To help you get started we've provided some classification tutorials as well as a quick tutorial on how to scrape images from google image search to help you build your training data set. Deep learning is allowed, but do remember that the hack is time limited!
The Santa Hack was inspired by this blog.
Please make sure you read the code of conduct for the hackathon (and all hackathons). As well as reading it, you are also welcome to re-use it. Ours is adapted from one produced by the RAS, but Github also provides a set of standard templates for collaborative coding projects that you might find useful.
Not at the hackathon, but want to test your code-building skills? Feel free!
> git clone https://github.com/as595/4IRSantaHack.git
Then make sure you have the right Python libraries for the tutorials. They can all be installed using pip and the requirements.txt file in the repo:
> pip install -r requirements.txt
The 4IR CDT is a partnership between:
GIF credit: AnimatedImages.org