In today’s post I’d like to introduce you to Azure Computer Vision API and tell you about some of its powerful functionality and what you can use it for.
- Analyzing an image. When we provide an image to the API and it can provide us a description of the image, as well as list out objects it sees in an image (people, cars, buildings, etc.). With people it can recognize whether there are faces in the image, gender of the faces and will even attempt to determine the age of the people.
- Optical Character Recognition (OCR) – in simple terms, reading text from an image.
- Image Recognition – The Computer Vision API has a library of about 200,000 celebrity images and 9000 landmarks, so we can use it to identify whether a famous person or landmark is included in the picture.
- Analyze and read handwritten text. A bit like OCR, but it will read handwritten text and convert that into character strings that we can use.
You can go to the Azure Computer Vision site and try this yourself. I’ve got a few examples below of some things that I did:
- When on the website, I had a photo of Wrigley Field saved on my laptop and I browsed to the file location and sent it up to the API piece that analyzes text. In this screen shot you’ll see the results of how it recognized and read the text on the picture. So, it recognized that it says Wrigley Field, home of the Chicago Cubs, etc. In any example I’ve done I’ve found it’s pretty accurate at reading text.
- Next, I chose a picture of a familiar landmark, the Statue of Liberty, taken from my wife’s phone and passed it through the landmark recognition section on the trial page. You can see in the resulting JSON that it identified the landmark. This API is also the one that will tell you whether there are celebrities in the picture and who those celebrities are. Here it correctly identified that there are no celebrities in this one.
- Another cool thing about the Image Recognition API is that it will give you a percentage of confidence, how confident is it that it made a correct match. So, you’ll see that indicated here as well.
- My last example, I uploaded a picture of a famous person and the API correctly determined that it’s Jackie Robinson with a very high percentage of confidence. It also understood that there are no landmarks in the picture.
I did these just playing around on the website with the tooling available there. But the website also provides sample code snippets for how you can call this API from your application, so a handy reference there.
Another powerful application of this API is with PowerApps. With PowerApps you can use the phone operation on your camera, take a picture and use a PowerApps to call an API and describe that image.
I hope this gave you some ideas of how the Azure Computer Vision API works and that you’ll go to the website and give it a try.
Need further help? Our expert team and solution offerings can help your business with any Azure product or service, including Managed Services offerings. Contact us at 888-8AZURE or [email protected].