Using Temporal Convolutional Networks to estimate ball possession in soccer games. (1st August 2023)

• Record Type: Journal Article
• Authors: Borghesi, Matteo; Costa, Lorenzo Dusty; Morra, Lia; Lamberti, Fabrizio
• Is Part Of: Expert systems with applications. Volume 223(2023)
• Publication Date: 2023-08-01
• Subjects: Sport analytics -- Deep learning -- Tracking data -- Ball possession -- Game state -- Temporal Convolutional Networks; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2023.119780
• Languages: English
• ISSNs: 0957-4174

Abstract

Abstract: The use of tracking data in the field of sport analytics has increased in the last years as a starting point for in-depth tactical analyses. This work investigates the use of Temporal Convolutional Networks (TCNs), a powerful architecture for sequential data analysis, to extract ball possession information from tracking data. This task is a crucial step for many tactical analyses and is nowadays carried out manually by a human operator in the stadium, which is costly, difficult to implement, and prone to errors. In this work, several classification approaches are explored to classify the game state as dead, ball owned by the home team, or by the away team: as a single-branch, ternary prediction, or as two binary predictions, first detecting whether the game is dead or alive and then which team owns the ball. TCNs are exploited to create independent trajectory embeddings from tracking data of each object; since there is no semantic ordering among the tracked objects, we investigate different permutation-invariant layers to combine the embeddings, namely, an element-wise sum over the embeddings, a self-attention module, and the use of 2D convolutions. Performance evaluation on tracking data from professional soccer games shows that the proposed method outperforms state-of-the-art rule-based methods, achieving 86.2% accuracy in possession estimation (＋7.3% compared to the state of the art) and 89.2% accuracy in dead-alive classification (＋33.2% compared to the state of the art). Extensive ablation studies were conducted to investigate how different input data concur to the final prediction. Highlights: Neural networks can be applied to estimate the state of a football game. Temporal convolutional networks are well-suited to handle spatio-temporal data. Permutation-invariant operators extract team-level features from trajectories. The best possession estimation is given by a two-network model with self-attention. A subset of trajectories is sufficient for successful game-state estimation.