AI Paper Atlas

Research map / Video / Action recognition

A Semantic and Motion-Aware Spatiotemporal Transformer Network for Action Detection

Matthew Korban, Peter A. Youngs, Scott T. Acton

TPAMI 2024 · Video › Action recognition · approach: Spatiotemporal CNNs for action recognition · 27 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add motion enhancement to motion-aware 2D positional encoding (ablation)SMAST · AVA 2.1mAP@0.5 (%): 31.0 -> 33.1Improved
use motion-aware 2D PE without motion enhancement vs standard 1D PE (ablation)SMAST · AVA 2.1mAP@0.5 (%): 30.9 -> 31.0No clear effect
add inter-feature attention A_GM to A_GG+A_MM (ablation)SMAST · AVA 2.1mAP@0.5 (%): 28.0 -> 31.8Improved
add inter-feature attention A_MG to A_GG+A_MM (ablation)SMAST · AVA 2.1mAP@0.5 (%): 28.0 -> 30.9Improved
add both A_GM and A_MG (full multi-feature attention) (ablation)SMAST · AVA 2.1mAP@0.5 (%): 28.0 -> 33.1Improved
use inter-feature attention alone (ablation)SMAST · AVA 2.1mAP@0.5 (%): 26.5 -> 24.4Hurt
use selective attention (Q persons, K persons+objects) instead of self-attention (ablation)SMAST · AVA 2.1mAP@0.5 (%): 31.8 -> 33.1Improved
restrict self-attention to objects only vs persons+objects (ablation)SMAST · AVA 2.1mAP@0.5 (%): 31.8 -> 24.3Hurt
use sequence-based temporal attention instead of efficient temporal attention (ablation)SMAST · AVA 2.1mAP@0.5 (%): 30.9 -> 32.5Improved
add sequence-based temporal positional encoding PTPE (ablation)SMAST · AVA 2.1mAP@0.5 (%): 32.5 -> 33.1Improved
efficient temporal attention instead of standard (ablation)SMAST · AVA 2.1mAP@0.5 (%): 30.4 -> 30.9No clear effect
plug multi-feature attention into AIAAIA · AVA 2.1mAP@0.5 (%): 31.2 -> 32.4Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy