If there’s one thing freshmen Avinash Jain and Joe Zou love almost as much as computer science (CS), it’s dance. So they were thrilled to join the University of California, Berkeley’s AFX Dance group in September 2018.
“Joining a college dance team was both exhilarating and tiring,” Jain says. “We had a flurry of new songs and pieces to practice. No matter how much effort we put in, it was incredibly difficult to accurately identify what specific moves we were performing wrong, and how we compared to the actual choreographer teaching the piece.” That gave Jain and Zou an idea.
They teamed up with fellow computer science students and dance enthusiasts Andi Gu and Nikhil Mandava. The four began developing an application to help dancers see exactly what moves were out of sync with the choreography and how they could improve them.
“We wanted to create a web platform where users could upload and view their own video and compare it side-by-side with their choreographer’s video,” Jain says. “We needed a server to host and render our entire web platform, a database for users to upload, store, and access video files, [and to] manage the users’ videos and compare them post-processing for easy analysis. In both of these scenarios, we came to Google Cloud to see what products we could use like Firebase Realtime Database, Cloud Storage, and App Engine.”
Using a body detection library for real-time multi-person keypoint detection, they began developing algorithms to analyze dancers’ movements. A few weeks later, they debuted their SyncUp app at the Bay Area collegiate hackathon Cal Hacks 5.0 in November 2018.