Welcome to the Lightning Pose documentation

Lightning Pose is an open source deep-learning package for animal pose estimation (preprint). The framework is based on Pytorch Lightning and supports accelerated training on unlabeled videos using NVIDIA DALI. Models can be evaluated with TensorBoard, FiftyOne, and Streamlit.

If you would like to try out Lightning Pose, we provide a Google Colab notebook that steps through the process of training and evaluating a model on an example dataset - no data labeling or software installation required!

We also provide a browser-based GUI that supports the full life cycle of a pose estimation project, from data annotation to model training to diagnostic visualizations.

Contents:

• Installation
  • Method 1: pip package
  • Method 2: conda from source
  • Method 3: Lightning Studio
  • Docker users
• User guide: the basics
  • Organizing your data
  • The configuration file
  • Training
  • Inference
  • Evaluation
• User guide: advanced
  • Unsupervised losses
  • Temporal Context Network
  • Multiview: mirrored or fused frames
  • Multiview: separate data streams
• Developer guide
  • Add a loss
  • Add a model
• FAQs
• Lightning Pose API

Indices and tables

• Index

• Module Index

• Search Page