-
Book Overview & Buying
-
Table Of Contents
-
Feedback & Rating

Computer Vision on AWS
By :

Developing a custom ML model to analyze images is a significant undertaking that requires tremendous time, ML expertise, and resources. Additionally, it generally requires thousands of hand-labeled images to provide the model with enough data to accurately make decisions. It would take months to gather this data and typically requires large teams of human labelers to prepare it for use in ML.
With Amazon Rekognition Custom Labels, you can offload this heavy lifting to the service. Custom Labels builds off of Amazon Rekognition’s existing capabilities (as explained in Chapter 2), using transfer learning (TL). Instead of you needing to provide thousands of images, you can take a small set of images (typically around 100-200 images) for each label to train a model. If your images are already labeled, you can directly import them into Custom Labels. If not, you can use Custom Labels’ built-in labeling interface or use SageMaker...