Skip to content

Status: Needs Review

This page has not been reviewed for accuracy and completeness. Content may be outdated or contain errors.

Pipeline & Graph API

Overview

Pipeline and graph construction in CUVIS.AI is handled through configuration files and the PyTorch Lightning framework. The cuvis_ai package provides nodes that can be composed into pipelines.

For information on building and configuring pipelines, see the guides below.

Pipeline Construction

Pipeline construction is done through:

  • YAML Configuration: Define pipelines declaratively using YAML config files
  • Python API: Build pipelines programmatically using PyTorch Lightning modules
  • Node Composition: Compose nodes from the cuvis_ai package into processing graphs

Node API Reference

For the complete API of available nodes that can be used in pipelines, see:

  • Nodes API - All available node implementations