SYSTEM AND METHOD FOR GENERATING TRACKABLE VIDEO FRAMES FROM BROADCAST VIDEO
A system and method of generating trackable frames from a broadcast video feed are provided 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 generates a set of frames for classification using a principal component analysis model. The set of frames are a subset of the plurality of video frames. The computing system partitions each frame of the set of frames into a plurality of clusters. The computing system classifies each frame of the plurality of frames as trackable or untrackable. Trackable frames capture a unified view of the sporting event. The computing system compares each cluster to a predetermined threshold to determine whether each cluster comprises at least a threshold number of trackable frames. The computing system classifies each cluster that includes at least the threshold number of trackable frames as trackable.
1 - 20 . (canceled)
21 . A method for predicting a player track, the method comprising:
receiving, by a computing system, a plurality of trackable frames from a database, wherein the plurality of trackable frames capture a unified view of a sporting event;
generating, by the computing system, a plurality of data sets from the plurality of trackable frames;
calibrating, by the computing system, a camera in each of the plurality of trackable frames based on the plurality of data sets; and
predicting, by the computing system, a track for each player based on the plurality of data sets and the calibrated camera.
22 . The method of claim 21 , wherein calibrating the camera in each of the plurality of trackable frames based on the plurality of data sets further comprises:
identifying, by the computing system, a frame subset of the plurality of trackable frames that provide a clear image of a playing surface;
comparing, by the computing system, the frame subset to a plurality of playing surface templates that include a different camera perspective of the playing surface; and
based on the comparing, identifying, by the computing system, a frame of the frame subset that matches a playing surface template as a keyframe.
23 . The method of claim 22 , wherein comparing the frame subset to the plurality of playing surface templates includes utilizing a neural network to perform the comparing.
24 . The method of claim 22 , wherein calibrating the camera in each of the plurality of trackable frames based on the plurality of data sets further comprises:
generating, by the computing system, a homography matrix for each keyframe.
25 . The method of claim 24 , wherein calibrating the camera in each of the plurality of trackable frames based on the plurality of data sets further comprises:
calibrating, by the computing system, the camera based on the homography matrix.
26 . The method of claim 21 , wherein the plurality of data sets include player location information, ball location information, and portions of a court corresponding to the plurality of trackable frames.
27 . The method of claim 21 , the method further comprising:
displaying, by the computing system, a graphical representation corresponding to each predicted track on one or more graphical user interfaces.
28 . The method of claim 21 , wherein the track for each player includes a set of real-world coordinates on a playing surface.
29 . A computer system for predicting a player track, the computer system comprising:
a memory having processor-readable instructions stored therein; and
one or more processors configured to access the memory and execute the processor-readable instructions, which when executed by the one or more processors configures the one or more processors to perform a plurality of functions, including functions for:
receiving a plurality of trackable frames from a database, wherein the plurality of trackable frames capture a unified view of a sporting event;
generating a plurality of data sets from the plurality of trackable frames;
calibrating a camera in each of the plurality of trackable frames based on the plurality of data sets; and
predicting a track for each player based on the plurality of data sets and the calibrated camera.
30 . The computer system of claim 29 , wherein calibrating the camera in each of the plurality of trackable frames based on the plurality of data sets further comprises:
identifying a frame subset of the plurality of trackable frames that provide a clear image of a playing surface;
comparing the frame subset to a plurality of playing surface templates that include a different camera perspective of the playing surface; and
based on the comparing, identifying a frame of the frame subset that matches a playing surface template as a keyframe.
31 . The computer system of claim 30 , wherein comparing the frame subset to the plurality of playing surface templates includes utilizing a neural network to perform the comparing.
32 . The computer system of claim 30 , wherein calibrating the camera in each of the plurality of trackable frames based on the plurality of data sets further comprises:
generating a homography matrix for each keyframe.
33 . The computer system of claim 32 , wherein calibrating the camera in each of the plurality of trackable frames based on the plurality of data sets further comprises:
calibrating the camera based on the homography matrix.
34 . The computer system of claim 29 , wherein the plurality of data sets include player location information, ball location information, and portions of a court corresponding to the plurality of trackable frames.
35 . The computer system of claim 29 , the functions further comprising:
displaying a graphical representation corresponding to each predicted track on one or more graphical user interfaces.
36 . A non-transitory computer-readable medium containing instructions for generating a player tracking prediction, the instructions comprising:
receiving, by a computing system, a plurality of trackable frames from a database, wherein the plurality of trackable frames capture a unified view of a sporting event;
generating, by the computing system, a plurality of data sets from the plurality of trackable frames;
calibrating, by the computing system, a camera in each of the plurality of trackable frames based on the plurality of data sets; and
predicting, by the computing system, a track for each player based on the plurality of data sets and the calibrated camera.
37 . The non-transitory computer-readable medium of claim 36 , wherein calibrating the camera in each of the plurality of trackable frames based on the plurality of data sets further comprises:
identifying, by the computing system, a frame subset of the plurality of trackable frames that provide a clear image of a playing surface;
comparing, by the computing system, the frame subset to a plurality of playing surface templates that include a different camera perspective of the playing surface; and
based on the comparing, identifying, by the computing system, a frame of the frame subset that matches a playing surface template as a keyframe.
38 . The non-transitory computer-readable medium of claim 37 , wherein comparing the frame subset to the plurality of playing surface templates includes utilizing a neural network to perform the comparing.
39 . The non-transitory computer-readable medium of claim 37 , wherein calibrating the camera in each of the plurality of trackable frames based on the plurality of data sets further comprises:
generating, by the computing system, a homography matrix for each keyframe.
40 . The non-transitory computer-readable medium of claim 39 , wherein calibrating the camera in each of the plurality of trackable frames based on the plurality of data sets further comprises:
calibrating, by the computing system, the camera based on the homography matrix.