Google releases software to help machines see

3 Dec 2015

With just a few hundred lines of Python code, this demo robot can roam and identify objects, including smiling faces.

Google has released new APIs and algorithms to developers to enable them to put machine vision and image recognition capabilities into their applications.

Put simply, Google wants to make it easier to enable computer to see as humans do.

Today (3 December), Google released its Google Cloud Vision API to enable developers to create applications that use computers and phones to understand images, recognise faces, logos and characters, detect objects and understand their environment.

“The uses of Cloud Vision API are game changing to developers of all types of applications and we are very excited to see what happens next,” explained Ram Ramanathan, product manager for the Google Cloud Platform.

Ramanathan explained that platforms like TensorFlow have enabled models that can learn and predict the content of an image.

Machine vision for the masses

Google Cloud Vision API 2

“Our limited preview of Cloud Vision API encapsulates these sophisticated models as an easy-to-use REST API.

“Cloud Vision API quickly classifies images into thousands of categories for example ‘boat’, ‘lion’, ‘Eiffel Tower’ – detects faces with associated emotions, and recognises printed words in many languages.

“With Cloud Vision API, you can build metadata on your image catalogue, moderate offensive content, or enable new marketing scenarios through image sentiment analysis,” Ramanathan said.

For example, with facial detection, the software can detect when a face appears in photos and, as well as nose and mouth placement, can detect over eight attributes like joy and sorrow. He added that Google won’t store facial detection information on any Google server.

The machine vision APIs can also be used to identify popular natural manmade structures as well as identify product logos within an image.

He said that in future phases Google will add support for integrating the machine vision apps with Google Cloud Storage.

To show a simple example of the Vision API, Google built a Raspberry Pi based platform with just a few hundreds lines of Python code, calling the Vision API.

The demo robot can roam and identify objects, including smiling faces.

John Kennedy is a journalist who served as editor of Silicon Republic for 17 years

editorial@siliconrepublic.com