lightning-pose-app 2.3.0.2

Lightning Pose user interface for labeling, training, and visualization

Lightning Pose App

Web-based GUI for Lightning Pose — a semi-supervised pose estimation library for single- and multi-view animal tracking.

The app provides an end-to-end workflow:

• Project management — create and organize pose estimation projects
• Labeler — extract video frames and annotate keypoints
• Models — configure and launch model training, monitor progress in real time
• Inference — run trained models across video sessions
• Viewer — inspect predictions overlaid on video with per-keypoint controls

Installation

Install Lightning Pose and the app:

pip install lightning-pose lightning-pose-app

Usage

litpose run_app

Then open http://localhost:4200 in your browser.

Documentation

Full documentation, including installation guides and tutorials:

👉 https://lightning-pose.readthedocs.io

Requirements

• Linux or WSL (Windows Subsystem for Linux)
• NVIDIA GPU with CUDA 12+
• Python 3.10–3.12

Source

This package contains only the app server and compiled UI. The core modeling library lives at paninski-lab/lightning-pose.