Google Cloud Apps Admins

Build a Bionic Eye that extracts data from Photos with The Google Vision API.

Written by Christopher Alghini (Google Cloud Consultant) | February 20, 2018

Today, companies from all around the world are searching for reliable, innovative ways to learn more about their customers. As the marketplace for almost every industry becomes increasingly competitive, customer experience and satisfaction are becoming the primary differentiators for brands who want to set themselves apart from the crowd. Since the best way to discover what your audience needs is to learn more about them, it's no surprise that deep learning "artificial intelligence" services are growing increasingly popular.

As usual, Google and their innovative cloud platform have emerged at the very front of the AI battle, with a host of incredible APIs designed to provide insightful data about consumers. However, Google doesn't just give its customers a way to listen to their audience's words, or even read between the lines. Google Photos has also become the very first brand to deliver the power of image recognition features into the public eye. Using image classification and pattern matching algorithms, this technology is what makes it possible for us to search for photos on Google Images containing a certain object or person.


Going Open-Source with Cloud Vision

Like many of the features in the Google Cloud Platform, Google Vision is an open-source API. By choosing the "open source" format, Google has allowed developers who might not be able to design their own applications, to make the most of the pre-existing information within the Google framework. With the Vision API, you can design your very own application for understanding the content in images. What's more, you can even add image classification options into your existing apps by incorporating capabilities from the Google REST APIs.

If you're new to the developer environment, then it's worth noting that an "Application programming interface" or "API" is a kind of messenger that takes important requests on the behalf of your business app and delivers the response back to you from the Google Cloud. For example, the Vision API immediately takes requests to categorize image, detect objects, faces, and logos, and then displays them back to you in response. This makes it much easier for companies to gather useful information about their brand, and business from pictures that would otherwise be difficult to analyze and classify.


The Features of Google Vision

As an API for the Google Cloud platform, Google Vision taps into all the power and accuracy you would expect from a Google product. The Google Vision API gives today's app developers and technology innovators in the complex business environment the opportunity to understand the contextual content of images, by providing insights from fully-equipped machine learning models. The simple, yet effective REST API classifies the photos and visuals in your company server into a host of different categories, and then detects the individual faces and objects within those images.

Not only can Google Vision pick a face out of a crowded image, but it can also read the words that are printed on an image, to make it easier for you to build metadata on a catalog of visual content. What's more, with Google Cloud Vision, you can enhance your current marketing schemes, moderate content that users might find offensive, and enable sentiment analysis too. Just some of the features advertised with the Google Vision API include:

  • Offensive content detection: Instantly find and remove explicit photos from your catalog, like violent images or adult content.

  • Label detection: Find and sort your images into different categories based on the content within an image, ranging from animals to people, and objects.

  • Landmark detection: Track man-made and natural structures in an image, like Big Ben, or the Grand Canyon.

  • Logo detection: Find and categorize popular product logos in your images.

  • Web detection: Browse through image galleries across the internet to find similar visuals.

  • Face detection: Find various faces within the same image, along with important attributes like emotional state. This could help you to determine the overall emotion at a company event, for instance.

  • Optical character recognition: The Google API can find text written within an image, and help you search for important visuals based on the content you're looking for.

  • Image attributes: Find general attributes within the picture, such as dominant colors or items.

All of that, and you can also enjoy complete access to the REST API which allows you to request multiple types of annotation per image. Those visuals can be uploaded into the Google Cloud, or integrated with the Cloud Storage feature.


The Potential of Image Recognition

Image recognition with Google Vision is just another incredible way to take your business intelligence to the next level. The more technology like this continues to evolve, the more new and incredible possibilities emerge for the future of the world as we know it. For instance, with image recognition and the Internet of Things, it could be possible to open a door without a key or unlock a safe from facial recognition alone.

The benefits that Google lists for their Cloud Vision API include:

  • The Opportunity to Moderate Content: Because the Vision API can detect offensive or inappropriate content thanks to "SafeSearch", companies can remove violent or adult content from crowd-sourced images.

  • Extract Text: With the Google Optical Character Recognition (OCR) feature, you can detect text from within visuals, and assess it along with automatic language identification. Vision API also comes with support for a host of languages.

  • World-wide Intelligence: The Vision API uses all the in-depth knowledge that it's gathered from Google Image Search to find entries for images that include logos, celebrities, news events, and many other important aspects of a visual.

  • Image Insights: Companies can detect broad object sets in their images, including animals, flowers, people, and thousands of other possibilities from categories found within the Google category database. What's more, because it's an AI, or machine-learning based feature, it learns and improves over time as accuracy continues to be honed.

Exploring Image Recognition

The concept behind image recognition is that companies should be able to define complex visuals with words, as this makes it much easier for them to determine how a visual should be classified. The Cloud Vision API is just one of the many ways that Google is helping to inform the business world as we know it. For help tapping into the behind the scenes knowledge that your images can provide, reach out to Coolhead Tech!