Google has filed a patent that will help users choose the right video from the sea of videos that get uploaded on video sharing platforms pertaining to public events such as music concerts and sporting events. The patent also allows viewers to review these events from multiple camera angles, while also being able to determine the exact position of the video sources themselves — a feature that which could come handy on social media networks to find friends who are at the same event.
The patent application doesn’t restrict the purview of the patent to the company’s own YouTube video sharing portal, but also extends the scope to “popular video delivery systems” and social networks. Prima facie, that seems to allude to the implementation of the video search feature within the company’s search engine itself, but that also doesn’t preclude the possibility of bringing the capability to YouTube either.
Aims to Classify Videos Based on Content and Not Just Keywords
From what we could glean from the patent filing, Google has found a way to use a variety of factors such as computer vision and deep learning aided frame analysis and video metadata of uploaded videos to improve amateur videography of large spectacles such as music concerts and sporting events. The patent aims to help the users sift through multiple videos that flood social media networks and video sharing websites soon after such events, based purely on the quality of video content and not just by relying on the keywords associated with the video uploads.
The patent details how Google aims to collate such videos and classify them according to the different angles from which the various videos are shot. The patent application talks about how the system employs geometric interpretation to determine the position from which the videos will have been recorded. The company posits that such information related to the camera position will allow viewers to have a more informed choice as to which videos to view. A good example of such an implementation will allow users to know which amateur video recording of a concert has been done front and centre, which is definitely better compared to another video shot from the corners and further away from the stage.
Could Allow Users to Review Controversial Umpiring Decisions from Multiple Angles
The patent filing also provides a specific use case for sporting events, by illustrating how it could help viewers review controversial umpiring decisions, just to cite a use case. The patent explains how footage of different camera angles covering, say, the second base plate could come in handy during baseball games where a runner has been forced out by a fielder at the base. The patent also details utilising anchor points such as home bases in baseball, goal posts in rugby or football, the pitch itself in cricket, in addition to employing the natural roll, pitch, yaw, and panning of a handheld camera such as a smartphone to resolve camera location. The presence of more than two video sources will also allow google to triangulate the position of the cameramen with respect to aforementioned anchor points within the videos.
Interestingly, the positional data works both ways. According to the patent application, such improved camera position estimates could be used in a specialised “event-based video browser”. Google posits that such positional information associated with videos will allow viewers to see not only what other people were watching at the time, but also where the other people were when they were recording the event.
Could Potentially Help Law Enforcement with Mass Shootings and Terrorist Attacks
While the patent doesn’t mention this particular use case, it’s not uncommon to find social media networks and video sharing websites flooded with videos of accidents, natural calamities, mass shootings, terrorist attacks, and other untoward events. The capability to review these events from multiple crowd-sourced camera angles, along with the possibility of determining the exact position of the videographers themselves, could help law enforcement analyse everything from accidents to terrorist attacks that usually tend to get captured by multiple civilians on their smartphones.
The main takeaway from this patent would be Google’s capability to rely on actual video data and not just keywords to collate and provide better information on video coverage of sporting events, concerts, or even major events such as riots and natural calamities. This is in addition of providing viewers with easy access to multiple camera angles from applications ranging from reviewing controversial umpiring decisions to checking out accidents from different angles.
