DTN with DTPN using fast proposal linking for deformable tubes and DTRN with 3D conv nets claims SOTA action detection on UCF-Sports and AVA by modeling flexible shapes vs rigid cuboids.
Action tubelet detector for spatio- temporal action localization
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Deformable Tube Network for Action Detection in Videos
DTN with DTPN using fast proposal linking for deformable tubes and DTRN with 3D conv nets claims SOTA action detection on UCF-Sports and AVA by modeling flexible shapes vs rigid cuboids.