Abstract: The present embodiments relate to tracking player movements from a video broadcast and determining a defensive influence score from the tracked movements of the player. The present embodiments can implement one or more models to generate a defensive influence score that quantifies a defensive intensity of a player during the course of a game. The defensive influence score can include a frame-by-frame machine learning prediction that can be used to estimate the defensive pressure a player is having on another player during the course of the game. Additionally, the present embodiments can capture and estimate fitness metrics, such as sprints and efforts around detected plays such as pick-and-rolls and off-ball screens, which can be good proxies for player effort. Further, event detection outputs (both offensive and defensive metrics), can be used as features to estimate fitness metrics for the player (e.g., player load, sprints, jogs, etc.).

Abstract: One or more highlights of a video stream may be identified. The highlights may be segments of a video stream, such as a broadcast of a sporting event, that are of particular interest to one or more users. According to one method, at least a portion of the video stream may be stored. The portion of the video stream may be compared with templates of a template database to identify the one or more highlights. Each highlight may be a subset of the video stream that is deemed likely to match the one or more templates. The highlights, an identifier that identifies each of the highlights within the video stream, and/or metadata pertaining particularly to the one or more highlights may be stored to facilitate playback of the highlights for the users.

Abstract: A computing system retrieves ball-by-ball data for a plurality of sporting events. The computing system generates a trained neural network based on ball-by-ball data supplemented with ball-by-ball data with ball-by-ball match context features and personalized embeddings based on a batsman and a bowler for each delivery. The computing system receives a target batsman and a target bowler for a pitch to be delivered in a target event. The computing system identifies target ball-by-ball data for a window of pitches preceding the to be delivered pitch. The computing system retrieves historical ball-by-ball data for each of the target batsman and the target bowler. The computing system generates personalized embeddings for both the target batsman and the target bowler based on the historical ball-by-ball data. The computing system predicts a shot type for the pitch to be delivered based on the target ball-by-ball data and the personalized embeddings.

Abstract: A method and system of generating agent and actions prediction based on multi-agent tracking data are disclosed herein. A computing system retrieves tracking data from a data store. The computing system generates a trained neural network by generating a plurality of training data sets based on the tracking data by converting each frame of data into a matrix representation of the data contained in the frame and learning, by the neural network, a start frame and end frame of each action contained in the frame and its associated actor. The computing system receives target tracking data associated with an event. The target tracking data includes a plurality of actors and a plurality of actions. The computing system generates, via the trained neural network, a target start frame and a target end frame of each action identified in the tracking data and a corresponding actor.

Abstract: A method for generating coupled play, drive, and game outcome predictions for a sporting event, the method including: inputting features for a sporting event into an initial model, the input features including historical game data and in-game data; determining a predicted outcome for at least one upcoming play with the initial model; determining, using the input features and predicted outcome, a predicted probability of each team winning the sporting event; and determining a probability of success for an action in the at least one upcoming play of the sporting event.

Abstract: One or more highlights of a video stream may be identified. The highlights may be segments of a video stream, such as a broadcast of a sporting event, that are of particular interest to one or more users. According to one method, at least a portion of the video stream may be stored. The portion of the video stream may be compared with templates of a template database to identify the one or more highlights. Each highlight may be a subset of the video stream that is deemed likely to match the one or more templates. The highlights, an identifier that identifies each of the highlights within the video stream, and/or metadata pertaining particularly to the one or more highlights may be stored to facilitate playback of the highlights for the users.

Abstract: Disclosed techniques relate to improving predictions of machine learning models. In an example, a method involves receiving, from a machine learning model, predictions associated with a class of scenarios. The method includes identifying, in response to receiving the predictions, a subset of the class of scenarios that are beyond a threshold tolerance of accuracy. The method includes based on identifying the subset of the class of scenarios a training data set that includes emphasized event data from a plurality of historical sporting events. The method includes generating an updated machine learning model and deploying the updated machine learning model.

Abstract: A method of generating predictions for teams and players for each team associated with a sporting event, the method including: receiving one or more top-down predictions for the sporting event; providing the top-down predictions as one or more top-down feature vectors to a computing system; receiving, by the computing system, a second set of feature vectors comprising data for one or more players associated with one or more respective teams and data for one or more teams associated with a sporting event; inputting the one or more top-down feature vectors and second set of feature vectors into a transformer-based neural network; and generating, using the transformer-based neural network, one or more predictions for the sporting event based on the one or more top-down feature vectors and the second set of feature vectors.

Abstract: A method for tracking one or more individuals during a sporting event, the method including: receiving, as an input, a geospatial data of a sporting event; receiving, as an input, labeled event data of the sporting event; performing multi-object tracking of one or more agents of the geospatial data to determine one or more vectors; inputting the labeled event data and one or more vectors into a diffusion model; and determining, using the diffusion model, one or more trajectory sequences for the one or more agents.

Abstract: A method for tracking one or more individuals during a sporting event, the method including: receiving, as an input, a geospatial data of a sporting event; receiving, as an input, labeled event data of the sporting event; performing multi-object tracking of one or more agents of the geospatial data to determine one or more vectors; inputting the labeled event data and one or more vectors into a diffusion model; and determining, using the diffusion model, one or more trajectory sequences for the one or more agents.

Abstract: A method for generating predictions for teams and players associated with a sporting event using a transformer neural network, the method including: receiving a set of input features, the set of input features representing a set of players and teams within a match, each input within the set of input features being represented by a tensor; inputting the set of input features into a transformer neural network, the transformer neural network including: a set of embedding layers; transformer encoder layers; and fully connected layers; and generating, using the transformer neural network, a set of target metric predictions for the set of players and teams within the match.

Abstract: A system and method of predicting a team's formation on a playing surface are disclosed herein. A computing system retrieves one or more sets of event data for a plurality of events. Each set of event data corresponds to a segment of the event. A deep neural network, such as a mixture density network, learns to predict an optimal permutation of players in each segment of the event based on the one or more sets of event data. The deep neural network learns a distribution of players for each segment based on the corresponding event data and optimal permutation of players. The computing system generates a fully trained prediction model based on the learning. The computing system receives target event data corresponding to a target event. The computing system generates, via the trained prediction model, an expected position of each player based on the target event data.

Abstract: A system and method of calibrating a broadcast video feed are disclosed herein. A computing system retrieves a plurality of broadcast video feeds that include a plurality of video frames. The computing system generates a trained neural network, by generating a plurality of training data sets based on the broadcast video feed and learning, by the neural network, to generate a homography matrix for each frame of the plurality of frames. The computing system receives a target broadcast video feed for a target sporting event. The computing system partitions the target broadcast video feed into a plurality of target frames. The computing system generates for each target frame in the plurality of target frames, via the neural network, a target homography matrix. The computing system calibrates the target broadcast video feed by warping each target frame by a respective target homography matrix.

Abstract: A method of generating an outcome for a sporting event is disclosed herein. A computing system retrieves tracking data from a data store. The computing system generates a predictive model using a deep neural network. The one or more neural networks of the deep neural network generates one or more embeddings comprising team-specific information and agent-specific information based on the tracking data. The computing system selects, from the tracking data, one or more features related to a current context of the sporting event. The computing system learns, by the deep neural network, one or more likely outcomes of one or more sporting events. The computing system receives a pre-match lineup for the sporting event. The computing system generates, via the predictive model, a likely outcome of the sporting event based on historical information of each agent for the home team, each agent for the away team, and team-specific features.

Abstract: A system and method of calibrating moving cameras capturing a sporting event is disclosed herein. A computing system retrieves a broadcast video feed for a sporting event. The broadcast video feed includes a plurality of video frames. The computing system labels, via a neural network, components of a playing surface captured in each video frame. The computing system matches a subset of labeled video frames to a set of templates with various camera perspectives. The computing system fits a playing surface model to the set of labeled video frames that were matched to the set of templates. The computing system identifies camera motion in each video frame using an optical flow model. The computing system generates a homography matrix for each video frame based on the fitted playing surface model and camera motion. The computing system calibrates each camera based on the homography matrix generated for each video frame.

Abstract: Metadata for one or more highlights of a video stream may be extracted from one or more card images embedded in the video stream. The highlights may be segments of the video stream, such as a broadcast of a sporting event, that are of particular interest. According to one method, video frames of the video stream are stored. One or more information cards embedded in a decoded video frame may be detected by analyzing one or more predetermined video frame regions. Image segmentation, edge detection, and/or closed contour identification may then be performed on identified video frame region(s). Further processing may include obtaining a minimum rectangular perimeter area enclosing all remaining segments, which may then be further processed to determine precise boundaries of information card(s). The card image(s) may be analyzed to obtain metadata, which may be stored in association with at least one of the video frames.

Abstract: Customized highlight shows for sporting events, entertainment events, and/or the like, may be generated and presented. Source content with a representation of one or more events, such as sports events, may be obtained. A user may designate an attribute and/or a viewing length. A subset of the source content with highlights from within the source content may be selected such that the subset has the attribute and/or viewing length designated by the user. A customized highlight show including the subset may be generated and outputted to the user, for example, on a display screen. The attribute may include a team playing in the sporting events, a player playing in the sporting events, and/or a type of play occurring in the sporting events. The subset may optionally be generated based on teams and/or players in one or more fantasy sports teams.

Abstract: Customized highlight shows for sporting events, entertainment events, and/or the like, having, a narrative component are generated and presented. The events can be sporting events, entertainment events, and/or the like. For example, in the context of sporting events, a determination is made as to what types of sports, teams, leagues, players plays and/or the like are of interest to the user. A customized highlight show is then generated and presented, containing those specific portions of the sporting events that are likely to be of interest, arranged in a manner that is likely to be entertaining and interesting to the user and that presents' cohesive narrative.

Abstract: A computing system receives data for a game. The data includes at least one of tracking data or event data. Based on the data for the game, the computing system determines that an event has occurred within the game. Based on the determining, the computing system generates a graphic responsive to the event. The graphic includes insights related to the event. The computing system recommends an image relevant to the event based on metatags associated with the event. The computing system generates a visual element by merging the image and the graphic.

Abstract: A computing system identifies broadcast video data for a game. The computing system generates tracking data for the game from the broadcast video data using computer vision techniques. The tracking data includes coordinates of players during the game. The computing system generates optical character recognition data for the game from the broadcast video data by applying one or more optical character recognition techniques to each frame of the plurality of frames to extract score and time information from a scoreboard displayed in each frame. The computing system detects a plurality of events that occurred in the game by applying one or more machine learning techniques to the tracking data. The computing system receives play-by-play data for the game. The computing system generates enriched tracking data. The generating includes merging the play-by-play data with one or more of the tracking data, the optical character recognition data, and the plurality of events.