PaperScope
LIVE · 2026-09-29 05:40 UTC

Groupwise Selective State-Space Filtering for Accurate and Streaming Action Boundary Detection

Mustafa Bora Çelik

Latestcs.CLcs.LGcs.AIcs.CV
arXiv ID
2609.33400 v1
Category
Submitted
2026-09-27

Abstract

Action boundary detection partitions untrimmed video into intervals without assigning action classes. We present a boundary-detection adapter operating on pre-extracted video features, learning temporal representations via groupwise selective scans. Learned group fusion and temporal modeling convert these into transition scores, which are decoded into boundary timestamps. Trained with boundary-time supervision, the class-agnostic model is evaluated on Breakfast, GTEA, and 50Salads using temporal tolerances and bipartite matching, achieving boundary $F_1$ scores of 0.457, 0.622, and 0.611. A stateful variant enables feature-streaming inference with zero neural look-ahead, one-sample peak confirmation, and bounded memory. Downstream systems can subsequently assign s

Comment: 5 pages,3 figures

arXiv abs page · PDF · same-day batch