We’re leveraging the power of computer vision to make your typical running platformer game a truly exhilarating experience!


Our love for fitness and health! We are strong promoters of regular exercise. We just wanted to do our bit to make an otherwise boring cardio very enjoyable. :)

What it does:

We use the Pose-net, a Neural Network that can detect human pose and gait in real time with high accuracy to make a runner platforming game where you leap across platforms by actually performing leaps in the waking world!

How we built it:

We leveraged the power of the Pose-net to get real time feed of human body pose and gait. We also came up with neat algorithms for accurately detecting rotations, running and jumping movements. But due to lack of time, we couldn’t use all these statistics for our running platformer game, but instead used just the jump detection data, built a mechanism for the Pose-net to communicate with our game through a Node.js server and made the game as awesome as we possibly could within the provided timeframe.

Challenges we ran into:

We had a tough time making the Pose-Net model communicate with our Unity game. Even though we wanted to use our super-cool turn, jump and run detection algorithms to make a deliberate platformer, we faced many technical challenges in using stream input for our 2d Unity game and with the limited time given, we had to resort to using just the jump detection to make our game.

Accomplishments that we're proud of:

We are proud of the fact that we’ve pieced together a very enjoyable cardio routine for people of all ages. If only we had more time to learn to use Unity for game development, we could’ve made this game even more feature-rich!

What we learned:

A LOT! This was our first time building a video game on Unity and that means a lot to us video gamers. My mate here and I had a great time thinking of really cool ideas during idea jam before we decided on this project. We weren’t very aware of the big obstacles ahead in making different pieces work together in harmony, but we did our best and things have turned out great. My mate, who’s more experienced than me in Distributed Systems, taught me a whole lot about interprocess communications – a branch of computer science I always shied away from until now. In conclusion, we learnt to appreciate the power of big ideas, the joy of building cool stuff together and the fact that learnt is a drop and unknown is an ocean.

What's next:

Just Jump 2! (Although it wouldn’t just be about jumping)

Built with:

Node.js, Javascript, Unity, C#, Tensorflow

Prizes we're going for:

Misfit Shine 2

$100 Amazon Gift Cards

Call of Duty: Black OPS 4 (XBOX ONE)

Team Members

Shishir Verma, Ajay Shaan Shanmugam
View on Github