Google’s Vision AI API enriches news coverage

June 4, 2019

Less than a minute

Do you face difficulties in searching for particular news images residing in the open Web?

If so, try Google’s Vision AI API.

For instance, Google’s Vision AI API was used to analyse a total of 812 hours of television news from CNN, Fox News, MSNBC and its affiliates in April this year. The AI API was able to identify, according to subjects specified, 167,937 web entities to include ‘Donald Trump’, ‘Robert Mueller’ and ‘Cathedral Notre-Dame de Paris’.

Although the experiment of Google’s advanced API began on the day of Notre Dame’s fire, it was able to instantly label the event as ‘Notre-Dame de Paris fire’ and continued to annotate during the course of event, with the captions of those images found online.

Indeed, Google’s Vision AI API has the ability to crowdsource available images across the web and train the datasets to recognise as well as annotate images found.

Broadcasters can now utilise the Vision AI API to quickly check and analyse real-time videos and images posted on the web — and potentially enrich their breaking news coverage.

Source: Forbes