Google Reveals New Machine Learning API that Can Search Objects inside Video

March 10, 2017, By Sanjeev Ramachandran

Video search results spawned by Google and YouTube are often satisfying. But at least at some point would we have wished for a search algorithm that would ponder inside a video for specific objects. That’s possible with images right now thanks to the bunch of machine learning features, and it could soon be a reality for videos as well.

During the Cloud Next conference held in San Francisco, the Mountain View company has revealed its new machine learning API, called Video Intelligence API, which possess the ability to automatically identify objects in video and provide search results based on them.

With the new API, developers will be able to extract metadata from videos using the API to bring up the right portion of the video which has been queried for. Type in tiger or flower, and the Video Intelligence API will take you to that part of video where it appears. The API also flaunts the ability to detect the scene changes inside the videos.

Don’t expect to get these advanced search results in YouTube right now. Google has revealed that its new API is in private beta for developers, indicating that the early focus is to offer an enterprise solution for large organizations and consumer technology companies. They should also be having the videos stored in Google Cloud Storage to make use of the new API.

Besides the announcement of its new product, Google has also revealed its acquisition of Kaggle, which the company calls the web’s largest data science community with over 800,000 data experts dedicatedly engaged in learning data analytics and machine learning.

© 2008-2012 - All rights reserved | Privacy Policy