Reference

This page gives an overview of exposed MAITE packages and modules. These include:

• maite.protocols - structural types defining expectations for MAITE components and primitives in each AI problem

• maite.tasks - core test & evaluation procedures operating on components and primitives

• maite.interop - wrappers transforming third-party objects into MAITE components

• protocols (by AI problem)
  • image classification
  • object detection
  • multiobject tracking
  • generic

• tasks
  • evaluate
  • predict
  • evaluate_from_predictions
  • augment_dataloader

• interop
  • TMClassificationMetric
  • TMDetectionMetric
  • YoloObjectDetector

Tip

Check out the MAITE Vision explainer for a clear description of components, tasks, and primitives in the context of the MAITE package.

Warning

Contents of maite._internals is considered private. It contains functionality that is either only used internally, or considered fundamentally more experimental. Users should have no expectations of stable functionality from this subpackage.