We’ve build an extension for Sky. As there is so many video/series available nowadays and some of them aren’t the simplest, it’s hard to keep track of all the details all the time. Our concept aims to solve this problem by providing a data-driven screen application that tells you everything you need to know and extending the user experience.
What it does
The core feature is to provide helpful, real and relevant data based on what can be seen live in the video. Next to this, we added a comment system where users can post a comment during the video and read comments of others.
How we build it
Starting off with OpenCV, where we use face recognition to extract flat faces of the video. These faces are send to our Watson backend module where trained Watson classifiers will map the face to a name. Once a match is found, the Watson module will send the name to a scraper module which will gather all the data it can find on the name. After this, the data is given to the Android application and will be visualised so the user can enjoy the extra content.
Challenges we ran into
One challenge was to collect decent data and cleaning it to train the classifier successfully. Another one was boosting the face recognition to a level that it can be used.
Accomplishments that we’re proud of
We achieved exactly what we had in mind.
What’s next for SkyFly
The possibilities are endless. Next to the data and comments, the second screen app can be extended with numerous of features to aid the users experience.
Watson, OpenCV, Android, Sky