Estimating missing player locations in broadcast video feeds
Examples disclosed herein may estimate locations of players not visible in a sporting broadcast video. A prediction model may be generated based on a training data set of in-venue tracking data that includes locations of all players at all times and the corresponding broadcast tracking data that may not necessarily contain the locations of all players at all times. The prediction model may be based on an algorithmic logic (e.g., a spline regression) or machine learning model (e.g., k-nearest neighbor, deep neural network). The generated predicted model may be used to estimate the unknown locations of players in broadcast tracking based on the known locations.
1 . A method comprising:
receiving, by a computing system, broadcast video data corresponding to a sporting event and in-venue video data;
analyzing, by the computing system, the broadcast video data to determine whether the broadcast video data includes a known starting player location at a start of the possession and a known ending player location at an end of the possession;
generating, by the computing system, a prediction model based on the broadcast video data and the in-venue video data, wherein the generating includes selecting the prediction model type based on whether the broadcast video data include the known starting player location and the known ending player location;
receiving, by the computing system, video data with at least one missing player location corresponding to a possession of the sporting event;
estimating, by the computing system, via the prediction model, a player location corresponding to the at least one missing player location; and
outputting, by the computing system, the estimated player location to a graphical user interface of a device.
2 . The method of claim 1 , the method further comprising:
utilizing, by the computing system, a machine learning model to minimize an error rate between an estimated location corresponding to the broadcast video data and a location in the in-venue video data.
3 . The method of claim 1 , wherein the broadcast video data includes tracking video data generated from the broadcast video data.
4 . The method of claim 1 , wherein the in-venue video data includes one or more player locations during the sporting event.
5 . The method of claim 4 , the method further comprising:
generating, by the computing system, at least one statistic based on the one or more player locations and the estimated player location.
6 . The method of claim 1 ,
wherein the prediction model type includes at least one of: a spline model type, a deep neural network type, a machine-learning model type, an end-to-end deep neural network type, or a k-nearest neighbors type.
7 . The method of claim 1 , wherein the broadcast video includes live broadcast video stream data or recorded broadcast video data.
8 . A system comprising:
a processor; and
a memory having programming instructions stored thereon, which, when executed by the processor, causes the system to perform operations comprising:
receiving broadcast video data corresponding to a sporting event and in-venue video data;
analyzing the broadcast video data to determine whether the broadcast video data includes a known starting player location at a start of the possession and a known ending player location at an end of the possession;
generating a prediction model based on the broadcast video data and the in-venue video data, wherein the generating includes selecting the prediction model type based on whether the broadcast video data include the known starting player location and the known ending player location;
receiving video data with at least one missing player location corresponding to a possession of the sporting event;
estimating, via the prediction model, a player location corresponding to the at least one missing player location; and
outputting the estimated player location to a graphical user interface of a device.
9 . The system of claim 8 , the operations further comprising:
utilizing a machine learning model to minimize an error rate between an estimated location corresponding to the broadcast video data and a location in the in-venue video data.
10 . The system of claim 8 , wherein the broadcast video data includes tracking video data generated from the broadcast video data.
11 . The system of claim 8 , wherein the in-venue video data includes one or more player locations during the sporting event.
12 . The system of claim 11 , the operations further comprising:
generating at least one statistic based on the one or more player locations and the estimated player location.
13 . The system of claim 8 , wherein the prediction model type includes at least one of: a spline model type, a deep neural network type, a machine-learning model type, an end-to-end deep neural network type, or a k-nearest neighbors type.
14 . The system of claim 8 , wherein the broadcast video includes live broadcast video stream data or recorded broadcast video data.
15 . A non-transitory computer-readable medium storing computer-executable instructions which, when executed by at least one processor, cause the at least one processor to perform operations comprising:
receiving broadcast video data corresponding to a sporting event and in-venue video data;
analyzing the broadcast video data to determine whether the broadcast video data includes a known starting player location at a start of the possession and a known ending player location at an end of the possession;
generating a prediction model based on the broadcast video data and the in-venue video data, wherein the generating includes selecting the prediction model type based on whether the broadcast video data include the known starting player location and the known ending player location;
receiving video data with at least one missing player location corresponding to a possession of the sporting event;
estimating, via the prediction model, a player location corresponding to the at least one missing player location; and
outputting the estimated player location to a graphical user interface of a device.
16 . The non-transitory computer-readable medium of claim 15 , the operations further comprising:
utilizing a machine learning model to minimize an error rate between an estimated location corresponding to the broadcast video data and a location in the in-venue video data.
17 . The non-transitory computer-readable medium of claim 15 , wherein the broadcast video data includes tracking video data generated from the broadcast video data.