Summary
None
Readily available
Always exercise with care and pay attention to how your body feels during each movement. Immediately stop or modify any exercise that causes pain or discomfort. Participants should take frequent breaks, especially if they notice muscle fatigue or weakness, as performing many repetitive motions in a short period of time can put unnecessary strain on the body.
If you want a Project Idea with full instructions, please pick one without an asterisk (*) at the end of the title.
Abstract
Dance, especially K-pop dance, has become a significant part of pop culture. Even without dance classes, many people have been learning dance through copying choreography from dance videos and dance tutorials. K-pop routines are fun and expressive, but they can also be tricky. Moves are fast, precise, and often depend on small details, such as hand placement, angles, and timing. How do you know if you’re really matching the choreography?
One way developers have gamified the dance learning process is through creating games like Just Dance and Dance Dance Revolution (DDR). These games turn movement into a challenge: follow along, stay on beat, and earn points for accuracy. These games make practicing feel like playing. The problem is, you’re limited to the dances the game includes. What if you wanted to learn a brand new K-pop dance, TikTok trend, or even a dance from a movie?
What if you could make your own dance game? Imagine uploading any dance video you want, dancing along in front of a camera, and getting feedback–like a score, a combo, streak, or highlights showing where you were “off.” That would mean you could learn any dance you’re interested in, not just the songs built into a commercial game.
This kind of custom dance game is possible using pose estimation, a type of computer vision that detects key body points (such as shoulders, elbows, knees, and ankles) from images or video. With a pose estimation model such as YOLO (You Only Look Once), you can track a dancer’s movements frame by frame. Then, you can compare your pose to a reference choreography video and measure how closely you match the moves over time.
Can you build a dance game that recognizes your body position and scores your performance? Can you improve the scoring system to focus on timing, pose similarity, or specific “key moves” in the choreography? With pose estimation and a little creativity, you can turn dance practice into an interactive AI-powered game–and learn choreography a whole new way.
To get started with pose estimation, check out the projects Using AI to Detect Proper Exercise Form or Teaching Machines to Understand Sign Language. These projects can provide an easier introduction to pose estimation if you are just getting started. As you read them, think about how you could trade their ideas to dance: swap the “correct form” examples for short choreography clips and compare your keypoints to a reference dancer, and experience what “good” performance means (overall similarity and timing). You can also add your own feedback approach, like flagging specific body parts that are “off” (arms or legs) or highlighting a few signature moves.
Bibliography
- Ultralytics. Pose Estimation. Ultralytics YOLO Docs. Retrieved January 27, 2026.
Materials and Equipment
- Computer with Internet access
Ask an Expert
Careers
If you like this project, you might enjoy exploring these related careers:









