Posted On: Nov 18, 2021
Amazon Rekognition Custom Labels is an automated machine learning (AutoML) service that allows you to build custom computer vision models to detect objects and scenes specific to your business needs without the need of in-depth machine learning expertise. Starting today, we have updated the Amazon Rekognition Custom Labels console to introduce step-by-step directions on how to manage, train, and evaluate your custom models. This revamped guided experience makes it even easier for you to train your own computer vision models in four simple steps with just a few clicks.
Customers can manage their custom models with projects, which are a set of resources needed to build and train a model. Datasets are a collection of labelled images that are used to train the model. Previously, datasets were not directly associated to projects. With this update, datasets will now be associated to projects created, making it even easier for customers to manage their custom trained models.
Customers who have previously trained models will see no impact. Amazon Rekognition will automatically associate the dataset used to train the most recent model with the project the model belongs to. Previous datasets that were never used or used to train an older version of the model can still be used by associating them to a new project.
In addition, we have introduced seven new APIs to make it even easier for you to build and train computer vision models programmatically. With these new APIs, you can: (a) create, copy, or delete datasets, (b) list the contents and get details of the datasets, (c) modify datasets and auto-split them to create a test dataset. To learn more about these new APIs, please visit this section of our documentation guide.