GlowTrack

Description
GlowTrack is a fluorescence-based method for automatically generating large, labeled image datasets for training motion capture models.

This page contains data referenced in the GlowTrack paper.

Dataset Labeling method Train / test # of landmarks Link
Mouse diverse GlowTrack (serial) Train & test 1 download
Mouse challenge Manual Test 1 download
Human hand train GlowTrack (parallel, both automatic & manual landmark selection) Train 50 / 12 download
Human hand test Manual Test 12 download

Contact
Dan Butler, Molecular Neurobiology Lab (Azim Group), Salk Institute