PCL-Former, a three-tier transformer pipeline for temporal action localization, reports top average mAP on three benchmarks, but its evaluation protocol and reported margins are internally inconsistent.
Vivit: A video vision transformer
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Hierarchical Multi-Stage Transformer Architecture for Context-Aware Temporal Action Localization
PCL-Former, a three-tier transformer pipeline for temporal action localization, reports top average mAP on three benchmarks, but its evaluation protocol and reported margins are internally inconsistent.