Project Description

Our local cycling team runs a small Sweepstake during the Giro de Italia and Tour de France every year. It is quite popular among my friends and cycling fans.

The sweepstake is based in filling some papers to choose the cyclists you think will perform better on those races. So some small boxes must be checked. One of my friends takes care of passing this paper selection to a digital environment.

In order to improve the process, and mostly as a pet project, as I find computer vision really interesting, I want to work on system which captures images using my phone, send them to my server, and doing some image magic, find the selected boxes and store them in a database.

As a stretch goal of future improvements, I would like to display the work in a React application.

Goal for this Hackweek

Write the image rendering code to detect and store the selected cyclists.

Resources

Here the 1st version of the project: https://github.com/arbulu89/lokatza

Results

I have achieved implementing the 1st version of the project. Using my phone, I'm able to receive images using a 3rd party app, detect the cells of the sweepstake and find the checked boxes. I didn't have the time to polish the detection though. Depending on light conditions or the surface where the paper is on, false positives are quite common. Besides that the goal of adding a new frontend has been totally impossible.

I has been a nice learning experience for computer vision and openCV. I might most probably try to continue working in my spare time