Get started with Azure Custom Vision

Azure Custom Vision is an image recognition service that lets you build, deploy, and improve your own image identifier models. An image identifier applies labels (which represent classifications or objects) to images, according to their detected visual characteristics. Unlike the Computer Vision service, Custom Vision allows you to specify your own labels and train custom models to detect them

The Custom Vision service uses a machine learning algorithm to analyze images. You can develop, submit groups of images that feature and lack the characteristics in question. You label the images yourself at the time of submission. Then, the algorithm trains to this data and calculates its own accuracy by testing itself on those same images. Once you’ve trained the algorithm, you can test, retrain, and eventually use it in your image recognition app to classify images. You can also export the model itself for offline use.

By the end of this lab, you will be able to:

• Labeling for labeling Object Detection
• How to object detection with Azure Custom Vision

Scenario

Recognition for new and old currency is a key function of the paper currency sorter. How to discriminate unfitness banknotes which became rough and fuzzy, even be damaged is an important task in financial institution. Different from traditional fitness banknote recognition based on extracting feature manually.

We will use Azure Custom Vision to build a solution that can detect whether the paper currency in the pictures.

Task 1: Create Custom Vision

• Search “custom vision” using the searching bar, then click on Custom Vision
• Click Create to create custom vision
• Choose Both on create options
• Select your Subscriptions and Resource group that you created
• Choose East US as region and named the custom vision that you want
• Select the Free F0 as Training pricing tier
• Select the Free F0 as Prediction pricing tier
• Click on Review + create

• Click Create to create custom vision
• Wait for deployment success
• Choose Go to resource after the deployment is completed

Task 2: Train you model with object detection

• Select Custom Vision portal to direct to a new browser page
• Click SIGN IN (use the same account that you sign in to azure portal)

• Click NEW PROJECT to create a new project
• Please follow the details below:

- Named your project

- Choose the resource that you create

- Choose Object Detection as “Project Types”

- Choose General as “Domains”

• Choose Create project

• Click Add images to add image to train

• Choose all the image files from your desktop and upload them
• Wait the files uploading

• Click Done when uploading succeed

• Click images which are Untagged
• Choose any image

• Choose the area that you want to detect
• Give a tag to it (which is dollar in this demo)
• Tag all images we uploaded which need to train and repeat the previous two steps

• After tagging all images, we can see all images which are classified to Tagged section

• Select Quick Training for this lab
• Click on Train button

• Wait the model training

• After the model train successfully, choose Quick Test to test the model

• There’s a result output

Task 3: Publish your model

• Select Publish

• Name the module as Model name
• Select the Prediction resource
• Click Publish to deploy

• Select Prediction URL, you can deploy it in your service

Conclusion

Congratulations! We now have learned how to:

• Labeling for Labeling Object Detection
• Using Azure Custom Vision to detect object