Google has detailed new ways it is using artificial intelligence to improve enterprise products, and the most exciting among those details is its new Video Intelligence API. With this new technology, which is currently in a Private Beta according to Google, developers are able to search for specific objects within videos, a feature that could prove useful for surveillance videos, among other things.
The new Cloud Video Intelligence API, says Google, utilizes ‘powerful deep-learning models’ to search for user-specified objects (cats, humans, cars, etc) within video content, pinpointing moments within the video where these objects are located. For example, a long surveillance video of an empty parking lot can be quickly scanned through by having the feature pinpoint only parts in the video where a human is present.
According to Google, developers can specify both verbs and nouns when searching for content, meaning you could search for something like ‘drive’ just as well as something like ‘car.’ The API also supports contextual understanding for the moment when the desired objects appear within the video.
The company goes on to say that its new video API is targeted at consumer tech companies and large media organizations, and that it can help create media catalogs for video content, and also work to managed content from crowdsourced locations. The API can also be used by companies as part of their own video software projects.