Live stream processing and sensor-based measurement presentation therein

1 . A system comprising:

one or more processors; and

one or more memory storing instructions that, upon execution by the one or more processors, configure the system to:

determine, in-real time relative to players being on a field and during a live video stream showing the players on the field, first real-time data of a first player, the first real-time data generated based at least in part on a first sensor associated with the first player, the first real-time data generated during a play being performed by a second player;

determine, in real-time relative to the players being on the field and during the live video stream, second real-time data of the second player, the second real-time data generated during the play based at least in part on a second sensor associated with the second player;

determine first historical data associated with the first player;

determine second historical data associated with the second player;

generate, during the play, a first input to a first artificial intelligence model based at least in part on the first real-time data, the first historical data, the second real-time data, and the second historical data;

determine a first output of the first artificial intelligence model, the first output being based at least in part on the first input and indicating a prediction of an action to be performed by the first player relative to the second player during the play;

cause, based at least in part on the prediction, a first presentation of a first graphical user interface element in the live video stream during the play, the first graphical user interface element indicating that the action is predicted to be performed by the first player;

determine, after an end of the play, third data associated with the play;

generate a second input to a second artificial intelligence model based at least in part on the third data;

determine a second output of the second artificial intelligence model, the second output being based at least in part on the second input and indicating a difficulty of the play;

generate, based on the second output, a difficulty measurement for the second player; and

cause a second presentation of a second graphical user interface element in the live video stream, the second graphical user interface element indicating the difficulty measurement.

2 . The system of claim 1 , wherein the first input comprises a set of vectors indicating a ball position and, per player, a player statistic and at least one of: a velocity, an acceleration, a direction, a position, a position group, an indication of offense or defense, or a relative player position.

3 . The system of claim 1 , wherein the first output comprises a set of vectors associated with a team indicating, per player on the team, a classification of whether the player is predicted to perform the action or a likelihood of the classification, and wherein causing the first presentation comprises instructing a remote computer system configured to present the live video stream to show or hide the first graphical user interface element per player on the team.

4 . The system of claim 1 , wherein the second input indicates a first set of metrics during the play of the first player, a second set of metrics during the play of a third player on a same team as the second player, and a third set of properties of the play.

5 . A computer-implemented method comprising:

determining, in-real time relative to a first player being in a physical space, first real-time data of the first player, the first real-time data generated during a play in the physical space by at least using one or more sensors also present in the physical space;

generating, during the play, a first input to a first artificial intelligence model based at least in part on the first real-time data;

determining a first output of the first artificial intelligence model, the first output being based at least in part on the first input and indicating a prediction of an action to be performed by the first player during the play in the physical space;

repeating execution of the first artificial intelligence model during the play such as a plurality of outputs of the first artificial intelligence model is generated;

causing, based at least in part on the prediction, a first presentation of a first user interface element during the play, the first user interface element indicating the prediction, the first presentation updated during the play based at least in part on the plurality of outputs; and

determining a start and an end of the play, wherein the execution of the first artificial intelligence model is repeated at a predefined prediction rate between the start and the end such that the plurality of outputs of the first artificial intelligence model is generated at the predefined prediction rate.

6 . The computer-implemented method of claim 5 , further comprising:

determining, after an end of the play, second data associated with the play;

generating a second input to a second artificial intelligence model based at least in part on the second data;

determining a second output of the second artificial intelligence model, the second output being based at least in part on the second input and indicating a difficulty of the play;

generating, based at least in part on the second output, a difficulty measurement for a player that performed the play, the player being the first player or a second player; and

causing a second presentation of a second user interface element indicating the difficulty measurement.

7 . The computer-implemented method of claim 5 , further comprising:

determining, in real-time relative to players being in the physical space and during a live video stream showing players, second real-time data of a second player, the second real-time data generated during the play based at least in part on a sensor associated with the second player, wherein the second real-time data is determined by at least being received from the sensor or from a first server that receives sensor data of the sensor and generates the second real-time data from the sensor data;

determining first historical data associated with the first player; and

determining second historical data associated with the second player, wherein the first historical data and the second historical data are determined based at least in part on one or more application programming interface calls to a second server storing player statistical data, wherein the first input is generated further based at least in part on the first historical data, the second real-time data, and the second historical data, and wherein the first user interface element is caused to be presented in the live video stream.

8 . The computer-implemented method of claim 5 , wherein the first output is generated based at least in part on a first execution of the first artificial intelligence model during the play, wherein a second output of the first artificial intelligence model is generated based at least in part on a second execution of the first artificial intelligence model after the first execution and during the play, wherein the second output of the first artificial intelligence model indicates a change to the prediction.

9 . The computer-implemented method of claim 8 further comprising:

determining, from the first output, a first likelihood of the action;

determining that the first likelihood exceeds a first threshold, wherein the first user interface element is caused to be presented based at least in part on the first likelihood exceeding the first threshold;

determining, based at least in part on the first likelihood exceeding the first threshold, a second threshold to use in association with ceasing the first presentation prior to the end of the play;

determining, from the second output, a second likelihood of the action;

determining that the second likelihood is smaller than the second threshold, wherein the change indicates that the action is no longer predicted to be performed by the first player based at least in part on the second likelihood being smaller than the second threshold; and

causing the first presentation to cease.

10 . The computer-implemented method of claim 8 , wherein the change indicates that the action is no longer predicted to be performed by the first player, and wherein the computer-implemented method further comprises:

causing, after the second output is generated, the first presentation to continue for a predefined amount of time; and

causing, after the predefined amount of time and based at least in part on the change, the first presentation to cease.

11 . The computer-implemented method of claim 5 further comprising:

determining that the first output indicates that the action is also predicted for a second player during the play; and

suppressing an indication of the action from being presented in association with the second player.

12 . The computer-implemented method of claim 11 further comprising:

determining, based at least in part on statistical play data of the second player, a statistical measure for the second player performing the action, wherein the suppressing is based at least in part on the statistical measure.

13 . One or more computer-readable storage media storing instructions that, upon execution on a system, cause the system to perform operations comprising:

determining, after an end of a play by a first player that is present in a physical space, first data associated with the play in the physical space, the first data generated by at least using one or more sensors also present in the physical space;

generating a first input to a first artificial intelligence model based at least in part on the first data, the first artificial intelligence model trained based at least in part on historical play data and play difficulty data;

generating a second input to a second artificial intelligence model based at least in part on second real-time data of a second player, wherein the second artificial intelligence model is trained based at least in part on historical player data and ground truth labels indicating historical occurrences of play actions;

determining a first output of the first artificial intelligence model and a second output of the second artificial intelligence model, the first output being based at least in part on the first input and indicating a difficulty of the play, the second output being based at least in part on the second input and indicating a prediction of an action to be performed by the second player;

generating, based at least in part on the first output, a difficulty measurement for the first player; and

causing a first presentation of a first presentation element indicating the difficulty measurement and a second presentation of a second element indicating the prediction.

14 . The one or more computer-readable storage media of claim 13 storing further instructions that, upon execution on the system, cause the system to perform further operations comprising:

determining, in-real time relative to the second player being in the physical space, the second real-time data of the second player, the second real-time data generated during the play.

15 . The one or more computer-readable storage media of claim 14 , wherein the second input is expressed in a first dimensional space, wherein the second output is generated by at least:

generating first vectors by encoding, by the second artificial intelligence model, the second input into a second dimensional space having more dimensions than the first dimensional space;

combining, by the second artificial intelligence model, the first vectors into second vectors; and

classifying, by the second artificial intelligence model, each second vector of the second vectors as to whether the action is predicted based at least in part on the second vector.

16 . The one or more computer-readable storage media of claim 14 , wherein causing the second presentation comprises causing a graphical user interface element to be presented in a live video stream in association with a presentation of the second player in the live video stream, wherein the graphical user interface element is presented for a time duration until an end of the play or a change to the prediction.

17 . The one or more computer-readable storage media of claim 14 , wherein the second real-time data is determined by at least being received, by the system that is an on-premise system, from a sensor associated with the second player or from a first server that receives sensor data of the sensor and generates the second real-time data from the sensor data, and wherein causing the second presentation comprises instructing a remote video system to present the second presentation element.

18 . The one or more computer-readable storage media of claim 13 , wherein the play is a first play, and wherein the one or more computer-readable storage media store further instructions that, upon execution on the system, cause the system to perform further operations comprising:

determining, after an end of a second play by the first player, second data associated with the second play;

generating a second input to the first artificial intelligence model based at least in part on the second data;

generating, based at least in part on a second output of the first artificial intelligence model in response to the second input, an update to the difficulty measurement for the first player; and

causing a second presentation of a second presentation element indicating the update.

19 . The one or more computer-readable storage media of claim 18 , wherein the first presentation is presented after the end of the first play and before a start of the second play, and wherein the second presentation is presented after the end of the second play.

20 . The computer-implemented method of claim 5 , wherein the first real-time data corresponds to a sports event at the physical space, and wherein the first presentation is presented as a live video stream of the sports event to a device of a remote viewer of the sports event other than the first player.