no code implementations • ICCV 2017 • Timo Milbich, Miguel Bautista, Ekaterina Sutter, Bjorn Ommer
Without any manual annotation, the model learns a structured representation of postures and their temporal development.
Action Classification Retrieval +2