Inspiration
Pickleball is one of the fastest-growing sports in the U.S., but most players plateau quickly because they rarely get structured, data-driven feedback on their technique. We wanted to build something that makes high-quality coaching more accessible by using AI to analyze form and performance. StrokeSensei was created around the idea that players should be able to understand and improve their mechanics the same way professional athletes do with video and analytics.
What it does
StrokeSensei helps players improve their pickleball strokes using computer vision and AI-driven feedback. Users can upload or record gameplay, and the system identifies different stroke types, evaluates form, and tracks consistency over time. It also provides personalized coaching suggestions based on performance and allows users to practice against AI-generated opponents to simulate real match scenarios.
How we built it
We built StrokeSensei as a full-stack application combining computer vision, a backend analytics system, and an AI feedback layer. The vision pipeline processes video input to detect player movement and classify strokes. These outputs are stored in a database to track user progress across sessions. On top of that, we built an AI system that translates performance data into clear coaching advice. The frontend provides an interface for uploading footage, reviewing analysis, and tracking improvement over time.
Challenges we ran into
The hardest part was making stroke detection reliable across different environments. Variations in lighting, camera angles, and player speed made consistent classification difficult. Another challenge was reducing latency so feedback could feel responsive rather than delayed. We also spent time refining how the AI explains mistakes so that the feedback is useful and not overly technical or confusing.
Accomplishments that we're proud of
We were able to build an end-to-end system that connects computer vision, data tracking, and AI coaching into a single workflow. The stroke classification system performs reliably across multiple shot types, and the feedback engine produces actionable insights instead of generic suggestions. We’re also proud of the AI training mode, which gives users a way to practice against simulated gameplay scenarios.
What we learned
We gained a deeper understanding of real-time computer vision challenges, especially in fast-moving sports contexts. We also learned how important it is to bridge the gap between raw model outputs and user-friendly insights. On the product side, we learned how to design feedback systems that are both accurate and easy for non-technical users to understand and apply.
What’s next for StrokeSensei
Next, we plan to improve stroke recognition accuracy and expand support for more advanced techniques and edge cases. We also want to move toward real-time feedback during live play and introduce more interactive training modes, including multiplayer and competitive practice environments. Over time, the goal is to expand StrokeSensei into a broader AI coaching platform for multiple sports.
Log in or sign up for Devpost to join the conversation.