In order to train a machine learning model on images, the image dataset needs to be labeled with annotations. Annotations are labels that help the user identify objects of interest in an image.
Automatic:
Done by training an initial model on a dataset, having the technology detect semantic keywords automatically, and then using human effort to correct its output and then retrain the model.
Semi-Automatic:
The user annotates the image then subjects it to feature extraction and clustering algorithms. This method is done with the use of an annotation tool like makesese.ai or Viame. Semi-Automatic annotations are more efficient than manual annotation, more accurate than automatic, useful for dynamic databases but require frequent user interface refinements.
Manual:
Annotations require humans to label an image with a name or description in order to define it.
