Label Unification and Correspondence in 3D Annotation GUI in Web Browser Multi-view pose annotation GUI. No build system — pure vanilla JS served as static files.

Full documentation, tutorials, and user guides: (https://talmolab.github.io/luc3d-docs)

• index.html — Main SPA with all app logic in inline script
• pose-data.js — Data model (Skeleton, Camera, Instance, FrameGroup, InstanceGroup, Session)
• interaction.js — Mouse/keyboard interaction, hit testing, drag handling
• overlays.js — Canvas rendering for pose skeletons and overlays
• triangulation.js — DLT triangulation + reprojection math
• video.js — WebCodecs video decoding (OnDemandVideoDecoder)
• viewport3d.js — Three.js 3D visualization
• timeline.js — SLEAP-like timeline widget
• file-io.js — File loading, calibration parsing, export (TOML/JSON/SLP)
• demo-data.js — Demo skeleton and camera data
• styles.css — All styling

python3 -m http.server 8080 --bind 0.0.0.0
# App: http://localhost:8080/
# Tests: http://localhost:8080/tests/test-runner.html

https://talmolab.github.io/luc3d/

• Three.js 0.147
• mp4box.js
• All loaded via script tags in index.html

Browser-based tests in tests/test-runner.html. Open in browser to run.

• scripts/json_to_slp.py — Convert JSON export to SLEAP .slp format
• scripts/json_to_h5.py — Convert JSON export to HDF5 format
• Require: h5py, numpy