To participate in the initiative, each user must upload a selfie to the site. The system applies black and white ‘face paint' to the user's face, taking into account the position and orientation of the face. Each fan can share the photo generated by the system on social media using the dedicated hashtags.  

The most interesting aspect of the project was the use of Computer Vision techniques to apply the 'face paint'. To do this, the following approach was adopted:

• use of Open CV classifier for face recognition to ensure that the system only accepts photos containing one or more faces. The classifier also performs the task of identifying the ROI on which to apply the black and white pattern;
• use of the dlib predictor to identify the 68 landmarks or control points that characterize a face and that are useful for calculating orientation;
• use of scikit-image for deformation of the image with the black and white pattern to be applied based on the positioning of the landmarks calculated by the dlib predictor;
• application of the pattern obtained in the previous step on the initial selfie taken and uploaded by the user on the site.

All these operations were carried out by the system in parallel using an asynchronous task queue manager. This choice was made in order to distribute the computing load and ensure the image processed uniformly on all cloud infrastructure servers used.

The campaign had unprecedented success and in times of heavy traffic, it took 12 virtual instances on OpenStack to manage the load because of the massive participation.

After the victory against Barcelona, the campaign was replicated for all Champions League matches, including the 2017 league victory, when the black and white pattern was replaced with 'tricolor' makeup to represent the Italian national flag.