Cognitive Services – Custom Vision Service Paid Subscription Available

New Cognitive Services Custom Vision Service paid subscription announced.
This mean you can predict images using Custom Vision Service in pay-as-you-go style over free trial subscription.

With current (one and only) free subscription allows you to create max 20 of project (prediction model) and predict 1000/day, and now you can exceed those using paid subscription.

Microsoft Azure Blog > Announcing new milestones for Microsoft Cognitive Services vision and search services in Azure

Microsoft Azure Website > Cognitive Services Pricing - Custom Vision Service


How to crate Custom Vision Service Subscription from Azure Portal

Creating Custom Vision Service Subscription from Azure Portal, you can select each subscription for training (crate and manage prediction model(project, in Custom Vision Service word)) and prediction (predict images using created project), and also keys are individually produced.

How to crate new project using Custom Vision Service Training API

You can create, upload images as training data and manage projects using Training API
Custom Vision Service Training APIs

When create new project, request URL is like as this below using Create Project.

Please get "domainId" beforehand using GetDomains.

[Header] Training-key: YOUR_TRAINING_KEY


API Test console available in each API definition page clicking [Open Test API Console].

You can manage your newly created project using Custom Vision Service Portal, accessing from .

Don't forget training after uploading & tagging images. Train Project API also available.
[Header] Training-key: YOUR_TRAINING_KEY

How to predict using Custom Vision Service Prediction API

Prediction APIs are also available (as same as before) using created & trained Custom Vision Service project.

Custom Vision Service Prediction APIs

Using Predict Image API, here is request URL alike as below to predict online image (via URL)
[Header] Content-Type: application/json Prediction-key: YOUR_PREDICTION_KEY
[Body] { "Url": "" }

Comments (0)

Skip to main content