STDPose combines a Dynamic-Aware Mask, temporal feature and heatmap fusion, and a mutual information loss to improve video pose propagation and pose estimation on sparsely-labeled videos, achieving new state-of-the-art results on PoseTrack17, PoseTrack18, and PoseTrack21.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SpatioTemporal Learning for Human Pose Estimation in Sparsely-Labeled Videos
STDPose combines a Dynamic-Aware Mask, temporal feature and heatmap fusion, and a mutual information loss to improve video pose propagation and pose estimation on sparsely-labeled videos, achieving new state-of-the-art results on PoseTrack17, PoseTrack18, and PoseTrack21.