Inspiration

Baseball pitching analysis has traditionally relied on expensive motion-capture systems and specialized equipment. While professional teams and some elite programs can access these technologies, many amateur players, students, and youth athletes cannot afford them.

As someone involved in baseball and AI development, I wanted to solve this accessibility problem. My goal was to make useful pitching analysis available to more players using something they already have: a smartphone camera.

This idea led to Outpace Baseball, an AI-powered pitching analysis platform designed to help players understand their mechanics, improve their performance, and access data-driven coaching at a lower cost.

What It Does

Outpace Baseball analyzes pitching videos and extracts information about a pitcher's movements and mechanics.

Key features include:

  • AI-based motion analysis: Uses YOLOv11 Pose to detect human body keypoints from video.
  • Pitching mechanics analysis: Examines movements such as stride, elbow position, knee movement, and arm slot.
  • Pitch speed estimation: Analyzes video to estimate pitching velocity.
  • AI-powered coaching: Uses Amazon Bedrock to generate personalized feedback and suggestions based on pitching analysis.
  • Pitch history: Helps players review and compare their pitching performance over time.
  • Subscription model: Integrates PayPal to support a paid Pro experience and make advanced features accessible through a recurring subscription.

How We Built It

We developed Outpace Baseball using Python, Streamlit, OpenCV, and YOLOv11 Pose.

The video analysis pipeline uses OpenCV to process uploaded pitching videos, while YOLOv11 Pose detects body keypoints to support biomechanical analysis.

We also integrated Amazon Web Services:

  • Amazon S3: Stores uploaded video data.
  • Amazon Bedrock: Generates AI-powered pitching feedback and coaching suggestions.

For monetization, we implemented a PayPal subscription integration using the PayPal Developer API, with a Sandbox environment for development and testing. FastAPI handles the backend subscription workflow, while Streamlit provides the user interface.

Our objective was to combine accessible computer vision, generative AI, cloud infrastructure, and digital payments into one practical baseball analytics platform.

Challenges We Faced

One of our biggest challenges was estimating pitching performance accurately from ordinary video.

Unlike professional motion-capture systems, smartphone videos can vary in camera angle, distance, lighting, and recording quality. These differences make consistent motion analysis difficult.

We worked on improving the video-processing pipeline, extracting body keypoints, and developing analysis methods that can provide useful feedback without requiring expensive specialized equipment.

Another challenge was integrating multiple technologies into one system. Connecting the video analysis pipeline, AI coaching, cloud storage, and PayPal subscription workflow required careful backend development and testing.

What We Learned

Through this project, we learned how to combine computer vision, pose estimation, generative AI, cloud services, and payment infrastructure into a practical application.

We also learned that building an AI product is not just about developing a model. It is equally important to make the technology accessible, create a usable experience, and establish a sustainable business model.

What's Next

Our next steps are to improve the accuracy of pitching analysis, collect more pitching data, and develop a smartphone application so that players can record and analyze their pitching motions more easily.

Ultimately, we want Outpace Baseball to make data-driven coaching accessible to amateur and youth baseball players around the world, regardless of their financial resources or access to professional facilities.

Built With

Share this project:

Updates

Submission history