The year 2024 has marked a turning point in Driblab’s evolution as one of the most prominent companies in advanced football data analysis. Driven by a relentless pursuit to meet the increasingly demanding and precise needs of professional football, Driblab has continuously improved all of its products, introducing new features and cutting-edge developments. Alongside our comprehensive event data coverage—the most extensive on the market with over 200 competitions—and our consulting service, widely recognized by high-value clients, we have added products such as Driblab HUB, a management and coordination platform for scouting departments of any size and scope, and Capture, an on-demand match data service offering advanced statistics to academies, women’s football teams, and competitions underserved by other providers.
Now, as 2024 comes to an end, Driblab embarks on a new era, marked by a rebranded identity and revamped website, serving as the launchpad for the exciting developments we are about to share. As clubs and organizations have become increasingly aware of the importance of data, Driblab has focused on deepening its analysis of every aspect of the game, aiming to provide the market with a complete understanding of what happens in every fraction of a second during a match. To achieve this mission, Driblab now boasts its own tracking system, an advanced optical tracking technology powered by computer vision, integrated into our product range for comprehensive football match analysis.
What is a tracking system?
Driblab’s tracking system involves a data capture and analysis process that detects and tracks every movement of players and the ball on the field. It uses information collected from cameras and computer vision algorithms, generating a digital replica of the pitch in real-time. This allows the precise tracking of player positions and movements with centimeter-level accuracy. The process consists of several stages to ensure comprehensive coverage and optimal precision during every second of the match.
- Scene Recognition
Scene recognition focuses on identifying various moments and scenarios within the game, distinguishing between relevant and irrelevant camera views. For example, certain camera angles, such as crowd shots or replays, do not contribute to data extraction for gameplay and are filtered out by the system.
- Line Recognition
Identifying the field’s lines is essential for establishing a precise spatial reference. During this phase, the system detects demarcation lines (e.g., goal lines, penalty areas, sidelines) to create a reference framework known as homography. This framework enables the exact positioning of players and the ball on a digital map with centimeter-level precision.
- Team Identification
At this stage, the system distinguishes and classifies players by team, typically using visual cues such as kit colors. This step is critical for structuring the tracking data and enabling team-specific analysis of game actions.
- Player Detection
Using artificial vision algorithms, the system identifies all players on the field, ensuring each player’s presence is recorded accurately at the right time and place.
- Player Identification
After detection, the system assigns each player a unique tag or marker, enabling continuous tracking of individual movements throughout the match for in-depth analysis.
- Player Tracking
Continuous tracking records every movement of the players in real time. This crucial phase ensures a steady flow of data, allowing for the study of player performance and positioning at any given moment in the game.
.gif)
- Data Imputation
The collected data is then stored in a database, prepared for analysis. This step organizes and processes the data so analysts at Driblab and its clients can derive precise insights into performance and strategies.
- Data collection for Invisible Players
To address instances where players are not visible on camera, Driblab uses a custom deep learning solution known as imputation modeling, which relies on GPS tracking data to predict the locations of all 22 players at all times. The model is trained to account for the movement history of all players when predicting individual positions.
This approach allows for more nuanced and precise predictions than simple mathematical interpolation. The model considers the relationships between players on the field and incorporates the ball’s position to predict hidden players’ movements effectively. Once trained, the model can predict player locations even when camera footage doesn’t capture all 22 players simultaneously.
Pushing the Boundaries of Analysis
The mission behind developing and implementing this system is to expand the scope of analysis accessible to any professional through data. This new product—and future developments stemming from it—pushes the limits of what data can achieve.
At Driblab, innovation is at the core of our identity. We have built our brand on three fundamental pillars: precision, prestige, and reliability. From the beginning, we set out to transform how professional football understands and utilizes data. Thanks to our cutting-edge approach, we have grown year after year, adapting to market demands and leading technological advancements in sports analysis.
Today, Driblab is much more than a data company; it is a complete ecosystem where every emerging need finds a tailored solution within our products. Our commitment to customization and excellence has allowed us to collaborate with over 100 clients worldwide, developing bespoke solutions and delivering analytical tools that exceed expectations.
At Driblab, we don’t just analyze the game; we dive deeper, understand it, and make it accessible, supporting every one of our clients with the highest level of rigor and expertise.
