Pose Pilot
The Project
Pose Pilot is a fitness app that integrates motion tracking technology with AI art, making exercise fun and easy. I built it to address a common problem, not enough exercise. This issue is often complicated due to costly gym memberships, intimidation, and lack of feedback from conventional video tutorials. Pose Pilot aims to be an easy and virtual workout buddy for the beginners, students, and busy people by offering a structured experience without requiring expensive wearables, sensors, or subscriptions. Functionally, Pose Pilot leverages computer vision and pose estimation via a webcam. It detects body keypoints to convert the actions into numerical data, calculating joint angles to evaluate posture. To ensure accuracy and prevent "cheating," it uses a state based logic system that requires users to return to a neutral position; idle, arms and legs in resting position, before registering a complete rep. I purposely designed it to keep players engaged through structured routines, countdown timers, and automatic progression to maintain workout pacing. Pose Pilot stands out for its privacy features; faces are automatically blurred, and no video data is ever saved or transmitted. However, the most challenging part is to handle real world variables like poor lighting and camera angles that caused keypoint jitter. I overcame this by utilizing relative geometry instead of absolute pixels and refining movement thresholds. Overall, Pose Pilot is an example to show how simple technology can help reduce barriers to physical activity. While it is not a replacement for professional training, it does give you a way to easily get moving as a beginner. By combining real time movement feedback with privacy and accessibility, Pose Pilot demonstrates how AI can make home workouts more interactive, affordable, and supportive.
