AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Design and Analysis of Efficient Attention in Transformers for Social Group Activity Recognition

Masato Tamura

IJCV 2024 · Architectures and efficiency › Attention and transformers · approach: Attention head and block redesign · 10 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
decompose group queries into location and shared layout queries (ablation)Ours · Volleyball group activity recognitionaccuracy: 95.1 -> 95.4Improved
decompose group queries into location and shared layout queries (ablation)Ours · Volleyball social group activity recognitionaccuracy: 63.7 -> 64.4Improved
add intra-group attention after inter-group attention (ablation)Ours · Volleyball group activity recognitionaccuracy: 94.1 -> 95.4Improved
add intra-group attention after inter-group attention (ablation)Ours · Volleyball social group activity recognitionaccuracy: 60.3 -> 64.4Improved
apply inter-group before intra-group attention instead of reverse order (ablation)Ours · Volleyball group activity recognitionaccuracy: 94.5 -> 95.4Improved
apply inter-group before intra-group attention instead of reverse order (ablation)Ours · Volleyball social group activity recognitionaccuracy: 63.2 -> 64.4Improved
use divided attention with 12 embeddings instead of naive attention with 6 (ablation)Ours · Volleyball group activity recognitionaccuracy: 95.7 -> 95.4Hurt
use divided attention with 12 embeddings instead of naive attention with 6 (ablation)Ours · Volleyball social group activity recognitionaccuracy: 61.6 -> 64.4Improved
use divided attention instead of naive attention on member identification (ablation)Ours · Volleyball group member identificationduplicated ratio (lower is better): 11.7 -> 9.94Improved
add joint individual action recognition training (ablation)Ours · Volleyball social group activity recognitionaccuracy: 62.1 -> 64.4Improved
add joint individual action recognition trainingEhsanpour et al. · Volleyball social group activity recognitionaccuracy: 57.6 -> 44.5Hurt
use multiple embeddings per group (Ours) instead of one embedding (previous method)Ours · Volleyball social group activity recognitionaccuracy: 60.6 -> 64.4Improved

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