Use Azure Custom Vision to make a Face Recognition Application

3 min readJun 21, 2021


Here I’ll show you how to use Azure’s custom vision portal to make a working application that will allow you to perform various tasks.

Go to the Azure Portal and make a Resource Group, in that Group we can start the service of Custom Vision.

As per your preference you can set the service. Then you will land on a page like this

From here you can first save up your API Key and endpoint since we’ll require that in the program that we write. Then we can go to the Custom Vision Portal. In this portal you can Create a new project

Once the project is created we will be in the Project Page

Now here you can add images and also mention the tags to each image. Once you are done adding images, you can hit that ⚙️Train button, which will ‘magically’ create the entire model for you — and within seconds you have a working Face Recognizing Model ready. After training is done don’t forget to publish the iteration and also note the name of the iteration you wish to use.

Now once the model is ready we can use it in our code. I’ll be sending a mail by one face and Start an EC2 Instance + Attach an EBS Volume by another face.

A Bad Quality GIF

So in the above GIF I showed the working of my program, which you can find here. As you can see the program first captures a few images and then through the Azure Custom Vision Endpoint it predicts the Face — The majority prediction is then taken into consideration and then the program runs a condition and checks who it is and what task needs to be done.

Another Bad Quality GIF

Now in the above GIF I showed my friend’s face which then by using the AWS SDK creates an ec2 instance and also an ebs volume and also attaches it to the instance.

I hope you all got to learn something out of this article. Go ahead and hit the 👏 if you liked the article.