How to Utilize AZURE Custom Vision AI to Power Business Solutions

By Thomas Owca| February 24, 2020
Artificial Intelligence has taken the world by storm and has expanded rapidly in recent years. Today, a growing number of industries are expanding their workflows and solving complex problems with AI technology.
The use of Computer Vision, a field of artificial intelligence, is one example of how businesses are improving both their workflows and product lines by training computers to interpret the visual world much more like a human being would.

Computer Vision can address a wide range of challenges across any verticals. In healthcare, Computer Vision can solve complex problems related to monitoring blood loss in real time, among other medical uses. It can also be found in the agricultural industry to make operations more efficient which leads to greater crop yields and, ultimately, more profit.

How does Custom Vision differ from Computer Vision?

While Computer Vision represents the ability to predict upon pre-uploaded images it has already seen before Computer Vision might be able to identify the Statue of Liberty in an image, but probably would not be able to tell the differences between two different models of cars. Custom Vision is like the name implies, a Computer Vision service with the ability to train your own AI models to your own unique and custom training data sets. It offers a compelling opportunity to solve complex business problems.

Creating a Custom Vision AI Model

Creating a custom visual AI model is relatively easy using Microsoft’s Azure Computer Vision and Custom Vision services. After creating a Cognitive Service resource in Azure Portal, you can use the Custom Vision Portal to upload training images, categorize/tag the images, train the model, view the performance metrics, and even test your latest AI model with your own test images. Doing so is relatively straight forward. In the Custom Vision portal, you would create a new AI project, choose your images as necessary and tag them appropriately. This might end up looking something like the following:





Clicking on the training button will allow you to run a ‘quick training’ on the current images. The deep neural network behind the scenes of Azure Custom Vision AI will analyze your images and return the performance metrics.





At this point, you would be able to use the ‘Quick Test’ option to call the prediction endpoint to see how your results fare.



Microsoft’s Azure Custom Vision AI is a worthy service to try. It can be taught to understand a variety of different objects or tags. It streamlines the availability of AI by offering something more accessible to a wider audience. Through the prediction API, your very own Artificial Intelligence models can be utilized with a variety of applications and programming languages.

If you’re interested in learning more about how you can apply Azure Custom Vision to address your unique business needs, or if you’d like to learn more about BDO Digital’s artificial intelligence, machine learning, and cognitive platform service offerings, contact us to speak to a team of Azure and AI experts.
      New call-to-action