eSports have been gaining traction ever since multiplayer video games started to emerge. One reason for the growth can be attributed to the availability of streaming platforms. Both professional and regular players like to share their experience in playing the game, and players around the world love to watch them. One popular streaming platform is Twitch and as of now, the most popular game watched is PlayerUnknown’s Battlegrounds (PUBG).

Making the application

PUBG is a multiplayer battle game where players loot for equipment and weapons and all the players fight each other; the last man standing wins, like in “The Hunger Games.” One problem viewers face in watching live streams of this game on Twitch is finding a stream that has the most action. The most exciting part of the game is when a small number of players are left and are forced to fight in an area that keeps getting smaller over time. Viewers would go back and forth to the Twitch streams page just to find that exciting stream. IBM developer advocates built a web application called PUBG Rotisserie, where the video shown is always the most action-packed live stream. The app was built with NFL RedZone in mind, where they show every touchdown in football games. The web app uses the Twitch API to fetch a list of live streams and with that list, it uses open source technologies to perform the computer vision algorithm. The web app watches the live streams in the background and gets a frame from that and sent to an optical character recognition (OCR) engine (from Tesseract OCR), deployed in a separate service.

Why is computer vision relevant?

Using computer vision to derive data from the stream made the task of switching from one stream to another easier. The app would always show the live stream that has the fewest remaining players. In that way, the user would always see some fighting from the streamer. The developers have made their code available on GitHub, along with links to the open source technologies used. A new developer journey titled “Implement computer vision for PlayerUnknown’s Battlegrounds live streams” will guide you through deploying this app. Computer vision can be greatly helpful in automating certain tasks. It gives the app the ability to understand data from images. PUBG Rotisserie uses OCR to get the number of players alive. The concept can easily be applied to more games in Twitch and can be applied in different streaming platforms.

Join The Discussion

Your email address will not be published. Required fields are marked *