Recognizing Team Formation in American Football

Atmosukarto, Indriyati; Ghanem, Bernard; Saadalla, M. M.; Ahuja, Narendra

doi:10.1007/978-3-319-09396-3_13

Cited by 5 publications

(5 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This will be overcome in future works by analyzing sequences of frames in place of single images. As mentioned previously, Atmosukarto et al explored video analysis of real gameplay footage [6]. Further advancing our research to perform video analysis as well as analysis of real gameplay footage will enable us to evaluate our method on the same dataset and thus validate our methods.…”

Section: Future Workmentioning

confidence: 71%

“…Atmosukarto et al analyzed a video and extracted the following three elements: (1) the frame in which the offensive team is lined up in a formation (with 95% accuracy), (2) the line of scrimmage (with 98% accuracy), and (3) the type of formation of the offensive team (with up to 67% accuracy) [6]. The third element they extract, the type of formation, is most similar to the work we present here.…”

Section: Related Workmentioning

confidence: 99%

“…Our study shows that our approach has great potential to perform well on real football footage. The dataset used by [6] is publicly available. We did not run our method on this dataset due to the difference in formation classifications and different input data requirements.…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

Automated Pre-Play Analysis of American Football Formations Using Deep Learning

et al. 2023

View full text Add to dashboard Cite

Annotation and analysis of sports videos is a time-consuming task that, once automated, will provide benefits to coaches, players, and spectators. American football, as the most watched sport in the United States, could especially benefit from this automation. Manual annotation and analysis of recorded videos of American football games is an inefficient and tedious process. Currently, most college football programs focus on annotating offensive formations to help them develop game plans for their upcoming games. As a first step to further research for this unique application, we use computer vision and deep learning to analyze an overhead image of a football play immediately before the play begins. This analysis consists of locating individual football players and labeling their position or roles, as well as identifying the formation of the offensive team. We obtain greater than 90% accuracy on both player detection and labeling, and 84.8% accuracy on formation identification. These results prove the feasibility of building a complete American football strategy analysis system using artificial intelligence. Collecting a larger dataset in real-world situations will enable further improvements. This would likewise enable American football teams to analyze game footage quickly.

show abstract

Section: Future Workmentioning

confidence: 71%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Automated Pre-Play Analysis of American Football Formations Using Deep Learning

et al. 2023

View full text Add to dashboard Cite

show abstract

“…Data presented in this study and the proposed terminology, similar to that used in other team-based invasion sports [ 5 , 22 , 52 ], have the potential to assist in the understanding and innovation of strategies and tactics used by AF teams. This could aid future AF research, enhance the knowledge of coaches at elite and improve fan and media discussions and understanding of the game.…”

Section: Discussionmentioning

confidence: 99%

Discovery and Characterisation of Forward Line Formations at Centre Bounces in the Australian Football League

Seakins,

Gastin,

Jackson

et al. 2023

Sensors

View full text Add to dashboard Cite

The extent of player formation usage and the characteristics of player arrangements are not well understood in Australian football, unlike other team-based invasion sports. Using player location data from all centre bounces in the 2021 Australian Football League season; this study described the spatial characteristics and roles of players in the forward line. Summary metrics indicated that teams differed in how spread out their forward players were (deviation away from the goal-to-goal axis and convex hull area) but were similar with regard to the centroid of player locations. Cluster analysis, along with visual inspection of player densities, clearly showed the presence of different repeated structures or formations used by teams. Teams also differed in their choice of player role combinations in forward lines at centre bounces. New terminology was proposed to describe the characteristics of forward line formations used in professional Australian Football.

show abstract

“…Similar to football, it is an outdoor sport. Several prior works have addressed video analysis in American football [6], [33], [34]. Most of these approaches use classical machine learning methods to model the interaction between players and the motion in the scene.…”

Section: A Sport Video Analysismentioning

confidence: 99%

Spotting Football Events Using Two-Stream Convolutional Neural Network and Dilated Recurrent Neural Network

2021

View full text Add to dashboard Cite

This paper addresses the problem of event detection and localization in long football (soccer) videos. Our key idea is that understanding the long-range dependencies between video frames is imperative for accurate event localization in long football videos. Additionally, proper event detection is not likely for fast movements in football videos without considering mid-range and short-range correlations between neighboring video frames. We argue that event spotting can be considerably improved by considering short-range to long-range frame dependencies in a unified architecture. To model long-range and midrange dependencies, we propose to use the dilated recurrent neural network (DilatedRNN) with long shortterm memory (LSTM) units, grounded on two-stream convolutional neural network (Two-stream CNN) features. While two-stream CNN extracts local spatiotemporal features necessary for fine-level details, the DilatedRNN makes the information obtained from distant frames available for the classifier and spotting algorithms. Evaluating our event spotting algorithm on the largest publicly available benchmark football dataset -SoccerNet-shows an accuracy improvement of 0.8% -13.6% compared to state of the art, and up to 30.1% accuracy gain in comparison to the baselines. We also investigate the contribution of each neural network component in spotting accuracy through an extensive ablation study.INDEX TERMS Deep learning, Dilated RNNs, Sport videos event detection, Two-stream CNNs.

show abstract

Recognizing Team Formation in American Football

Cited by 5 publications

References 43 publications

Automated Pre-Play Analysis of American Football Formations Using Deep Learning

Automated Pre-Play Analysis of American Football Formations Using Deep Learning

Discovery and Characterisation of Forward Line Formations at Centre Bounces in the Australian Football League

Spotting Football Events Using Two-Stream Convolutional Neural Network and Dilated Recurrent Neural Network

Contact Info

Product

Resources

About