Computer Vision on AWS

9

Labeling Data with Amazon SageMaker Ground Truth

Labeled data is key to developing an accurate and effective model using a supervised machine learning algorithm. Typically, machine learning practitioners spend 70% of their time labeling and managing data. It slows down innovation and increases cost. We saw in Chapter 3, and Chapter 7, how we needed high-quality labeled data to develop custom ML models. Although those services allowed a labeling interface in the built-in console, if you have a large number of images in your dataset, it can quickly become a monumental and cumbersome task to label them. You would either need to outsource the labeling responsibility or would need a solution to split the labeling workload across multiple labelers. ...