Image processing has already moved far beyond facial recognition and numbers. Today's algorithms can identify very specific parameters, evaluate contexts, and even analyze potential emotions of the people who appear in them. The demo of the Google Cloud Vision API is compelling in these respects and serves as a reminder of Mountain View's data collection capabilities.
The world was turned upside down when Facebook demonstrated the power of facial recognition in its tagging system. Despite the doubts and the complaints, at that moment we knew the genie had escaped the lamp, and the only option for algorithms was to improve at their tasks. Today they can invent faces and create entire landscapes with minimal prompts and age people, but what happens with existing photos? How good are they at interpreting "what they see"?
Try the Demo Yourself
To answer those questions, nothing beats the demo of the Google Cloud Vision API. It's a bit hidden, but you don't need to do anything weird to make it work: just drag an image into the box, or upload one from your hard drive manually. The processing phase takes a few seconds, and once finished, Cloud Vision divides the extracted information into several categories.
The first one is "Objects", and as its name suggests, it assigns a certainty percentage to each detected object. Then we have "Labels", with a series of descriptions that could be useful (or not) when indexing and classifying the image. "Web Entities" creates a list of entities linked to the content, with certainty values ranging from 0 to 1 (the closer to 1, the more "sure" the algorithm is), although the demo has a small bug here (the final number should never exceed 1).
The "Properties" section is even more interesting. On one hand, Cloud Vision reports which are the dominant colors of an image, separated by percentage, and each with its hex and RGB code. On the other, it offers cropping suggestions, anticipating how the final image would look depending on the aspect ratio.
The last tab is "Safe Search", a not minor aspect when distributing that photo on the web. Is its content delicate in some way? Is it a medical image or does it represent a violent scene? Cloud Vision will try to determine that and more. In short, give it a try with some images on the web, or why not, upload your own photographs. It will probably surprise you.
Official site: Click here
