Annotating data for ML
In this section, you learn why ML models need annotated data and not simply data! Furthermore, you will be introduced to a diverse set of annotation tools.
Learning from data
As humans, we learn differently from ML models. We just require implicit data annotation. However, ML models need explicit annotation of the data. For example, let’s say you want to train an ML model to classify cat and dog images; you cannot simply feed this model with many images of cats and dogs expecting the model to learn to differentiate between these two classes. Instead, you need to describe what each image is and then you can train your “cat-dog” classifier (see Figure 2.1).
Figure 2.1 – Training data for the cat-dog classifier
It should be noted that the amazing capabilities of ML models are closely related to and highly affected by the quality and quantity of the training data and ground truth. Generally, we need...
