Microsoft is working on a Shazam-like app for identifying video content

Microsoft might release a Shazam-like app for identifying video content, sometime in the future, as the the Redmond-based tech giant filed a patent describing a technology that envisions such an app.

The methods provided herein evaluate a set of conditions to determine whether or not a target video is an abridged version of a reference video.

The app could be used as a standalone product, but it could also work as a feature inside another platform, such as Microsoft Edge, for instance. It would work similarly to Google’s Copyright Match Tool, which is an app that automatically detects copyrighted video content.

This app patented by Microsoft would have the same goal, and it would give content creators, but also broadcasting companies, a viable way to verify copyrighted video materials.

Content creators, such as broadcasting companies, may need to detect copied video content as a part of their workflow. The detection of copied video content is also relevant for uncovering unauthorized copying of video content, e.g. the unauthorized uploading of content to Internet media platforms.

How would Microsoft identify video content?

Same as Google’s Copyright Match Tool, this app would make use of an existing database to identify the videos, and users would only need to provide the video, even in a recorded state, for the app to properly identify it.

Here’s how it would do it:

1. The method looks at a ‘target video’ and a ‘reference video’.
2. It checks three things: a) The target video doesn’t have all the shots from the reference video. b) The target video includes groups of shots that are also in the reference video. c) All the shots that are in both videos are in the same order.
3. If all these conditions are met, the method identifies the target video as a shortened version of the reference video.
4. The result of this check is then outputted.

When it comes to the existing database, this tool would make use of the Internet, and especially search engines, such as Google to identify the videos, so it won’t be dependent just on Google, as it would take the other Search Engines, such as Bing, into consideration.

Either way, the app allows for flexibility: users would be optionally asked to provide the video and the reference, but in case they can’t, the app will work with the existing data, automatically.

The target video and/or reference video may be selected by a user, or programmatically. In one particular example, the rarget video is selected by a user and there reference video is selected programmatically.

Microsoft does have one tool that identifies and verifies copyrighted material, but it’s only available in Microsoft 365, for apps, such as Word. It doesn’t have a tool that would do the same to video content, for instance, so this kind of app would be quite useful.

However, this feature might be released as a Copilot plugin, because the AI tool essentially, and theoretically, could already do the process of identifying video content. However, users must provide input to the AI tool, first, and for now, even if Copilot on Windows accepts images, such as screenshots, it still doesn’t accept video inputs.

But that might change.