AI Paper Atlas

Research map / Video / Action recognition

Relaxed Transformer Decoders for Direct Action Proposal Generation

Jing Tan, Jiaqi Tang, Limin Wang, Gangshan Wu

ICCV 2021 · Video › Action recognition · approach: Fully supervised temporal action detection · 213 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 boundary-attentive scaling (alpha=2) to features (ablation)RTD-Net · THUMOS14AR@50: 36.22 -> 41.52Improved
use MLP encoder instead of Transformer encoder (ablation)RTD-Net · THUMOS14AR@50: 33.69 -> 41.52Improved
use smaller input receptive field for Transformer encoder (ablation)RTD-Net with Transformer encoder · THUMOS14AR@50: 33.69 -> 36.01Improved
use smaller input receptive field (64 -> 16) for MLP encoder (ablation)RTD-Net · THUMOS14AR@50: 41.52 -> 39.56Hurt
add relaxed matcher (ablation)RTD-Net · ActivityNet-1.3AR@100: 71.88 -> 73.21Improved
add completeness head score to classification score (ablation)RTD-Net · THUMOS14AR@50: 41.08 -> 41.52Improved
add temporal positional embedding in MLP encoder (ablation)RTD-Net · THUMOS14AR@50: 37.07 -> 41.52Improved
add positional embedding at decoder attention (ablation)RTD-Net · THUMOS14AR@50: 37.43 -> 41.52Improved
use pre-enhancement MLP projection instead of post (ablation)RTD-Net · THUMOS14AR@50: 38.81 -> 41.52Improved
enhance by multiplication instead of concatenation (ablation)RTD-Net · THUMOS14AR@50: 37.92 -> 41.52Improved
increase decoder layers from 1 to 2 (ablation)RTD-Net · THUMOS14AR@50: 32.76 -> 38.92Improved
increase decoder layers from 6 to 9 (ablation)RTD-Net · THUMOS14AR@50: 41.52 -> 38.36Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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