UniSTFormer merges spatial and temporal attention for skeleton-based action recognition, achieving 91.0% on NTU-60 with 0.5M parameters and 1.1 GFLOPs.
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
-
UniSTFormer: Unified Spatio-Temporal Lightweight Transformer for Efficient Skeleton-Based Action Recognition
UniSTFormer merges spatial and temporal attention for skeleton-based action recognition, achieving 91.0% on NTU-60 with 0.5M parameters and 1.1 GFLOPs.