ASEA is a skeleton-based interaction recognition network that selects active joints via temporal-weighted L2 norms and applies cross-attention between individuals, achieving state-of-the-art accuracy on NTU-26, SBU, and Kinetics-10.
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
-
Learning Adaptive Node Selection with External Attention for Human Interaction Recognition
ASEA is a skeleton-based interaction recognition network that selects active joints via temporal-weighted L2 norms and applies cross-attention between individuals, achieving state-of-the-art accuracy on NTU-26, SBU, and Kinetics-10.