Posted On: Oct 27, 2020
Amazon Rekognition Custom Labels is an automated machine learning (AutoML) feature that allows customers to find objects and scenes in images unique to their business needs with a simple inference API. Customers can create a custom ML model simply by uploading labeled images. No ML expertise is required.
Today, we launched usability improvements to the Amazon Rekognition Custom Labels console that make it easier to import label lists from existing datasets, and to add, filter, and search labels within a dataset. Amazon Rekognition Custom Labels recommends that customers provide both a training and test dataset when creating a custom ML model. Previously, customers using the console would create a new test dataset and manually re-enter all the labels they just created in their training dataset. Often, this resulted in customers toggling back and forth between datasets to make sure both label lists matched, since typos, missing labels, or spelling differences (for example ‘awslogo’ vs. ‘AWS Logo’) could cause errors. Now, Amazon Rekognition Custom Labels enables customers to simply import a label list from an existing dataset when creating a new dataset, reducing manual work and chance of errors that could affect training.
As part of this launch, we also added new search, filter, and ‘quick add’ functionality to labels list in the dataset gallery view. Now, customers do not need to manually scroll their label list or enter the edit modal to add a new label. Instead, they can simply type in the label they are looking for and the list will automatically filter. To add a new label, customers can simply click on Create New Label without needing to enter a the Edit labels modal.
Get started with Amazon Rekognition Custom Labels today.