3STEP Sports, the nation’s largest youth sports organization and owner of The UCReport, asked Scrapbox to build a unique technology to measure football player speed from game film. With this powerful tool, coaches and recruiters at the nation's biggest college football programs can easily see how fast players move on the field, helping them make smarter recruiting decisions and find standout talent.
This project was structured as a phased, two-year engagement, with Phase One dedicated to launching a Minimum Viable Product (MVP) in Spring 2022. This initial release allowed us to bring essential functionality to market quickly, enabling immediate feedback, early sales, and user engagement. Building on this foundation, we moved into Phase Two, which culminated in the full Version 1 (V1) launch in Fall 2023.
The MVP achieved its goal, validating the platform’s value and informing our development path toward a fully realized V1. With the V1 launch, we introduced a more precise speed measurement algorithm, a user-friendly interface, and a strengthened backend to ensure scalability and consistent performance.
To meet the project’s technical demands, the Scrapbox team applied advanced computer vision and machine learning techniques. A careful balance between automated processes and manual inputs ensured the highest accuracy in speed metrics, resulting in a platform that is efficient, scalable, and intuitive. This final product provides reliable, actionable insights that empower our client’s decision-making and drive results.
You can learn more about The UCReport and our max speed platform online at theucreport.com.
To create a platform capable of delivering reliable and precise player metrics, the Scrapbox team utilized an innovative combination of Python-based computer vision libraries, custom neural network models, and a uniquely crafted user interface. This blend of technology and design allowed us to achieve high accuracy in speed measurement, setting a new standard for assessing player performance in youth sports.
One of the most significant challenges in developing this platform was the inconsistent quality of game footage, a unique hurdle in our project. While computer vision has been employed to calculate speed in controlled environments, the Scrapbox team faced the added complexity of working with raw, low-quality footage recorded by non-professionals—often parents capturing high school games. These videos frequently include poor lighting, shaky camera movements, and limited resolution, making it difficult to obtain precise measurements using traditional methods. Additionally, inconsistencies like poorly painted field markings and obstructed player views added to the challenge, setting this project apart from standard applications of computer vision in sports analytics.
To address these challenges, we developed custom preprocessing algorithms to enhance the clarity and usability of low-quality video. Our solution includes automated stabilization, noise reduction, and adaptive resolution enhancement to counteract the limitations of non-professional footage. These preprocessing techniques ensure that the critical details needed for accurate player tracking remain intact, even when the video quality is suboptimal. This processing pipeline allows us to maintain high standards of accuracy in speed measurement, regardless of the film's initial quality.
We also adapted our neural network models to operate effectively in environments with diverse lighting conditions, camera angles, and field inconsistencies. By training our convolutional neural networks (CNNs) on a dataset reflective of these challenges—including footage from various high school games across different regions—we refined our model's ability to recognize and track players accurately despite the irregularities in the video. This adaptation to real-world conditions is a key differentiator of our platform, enabling it to deliver consistent results where other computer vision-based systems might struggle.
The final product stands out for its resilience and reliability in processing high school football footage—an area where other platforms may falter. By tackling the unique difficulties of low-quality, user-generated video, our team at Scrapbox has created a platform that isn’t just powerful but practical for real-world application. This capability to handle non-ideal footage without compromising accuracy provides our client, The UCReport, with a distinct advantage in the competitive landscape of sports analytics.
This project’s success showcases Scrapbox’s expertise in merging cutting-edge technology with real-world usability, making advanced performance analytics accessible and reliable, even in less-than-perfect conditions. Through innovative computer vision, custom neural networks, and a creative user interface, our solution delivers highly actionable insights, empowering recruiters to make data-backed decisions confidently, regardless of the quality of game footage they work with.
%201.png)
