AI Paper Atlas

Research map / Graph learning / Knowledge and temporal graphs

Channel-Wise Topology Refinement Graph Convolution for Skeleton-Based Action Recognition

Yuxin Chen, Ziqi Zhang, Chunfeng Yuan, Bing Li, Ying Deng, Weiming Hu

ICCV 2021 · Graph learning › Knowledge and temporal graphs · approach: Graph reasoning for visual recognition · 1,185 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
replace all graph convolutions with CTR-GC (ablation)ST-GCN style baseline · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 83.4 -> 84.9Improved
replace only 2 graph convolutions with CTR-GC (ablation)baseline · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 83.4 -> 84.2Improved
add channel-specific correlations Q to a shared topology (ablation)CTR-GCN · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 83.7 -> 84.9Improved
add the shared topology A as a prior for the channel-specific correlations (ablation)CTR-GCN · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 84.0 -> 84.9Improved
use refinement instead of per-channel parameterized adjacency matrices (DC-GC*)graph convolution · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 84.0 -> 84.9Improved
use dynamic non-shared topology (CTR-GC) instead of static shared ST-GCgraph convolution · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 83.4 -> 84.9Improved
per-channel parameterized adjacency (DC-GC* vs DC-GC)DC-GC · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 84.2 -> 84.0Hurt
reduction rate r=16 instead of 8 (ablation)CTR-GCN · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 84.9 -> 84.5Hurt
Sigmoid instead of Tanh in correlation function (ablation)CTR-GCN · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 84.9 -> 84.6Hurt
ReLU instead of Tanh in correlation function (ablation)CTR-GCN · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 84.9 -> 84.8Hurt
use correlation function M2 or M1+ instead of M1 (ablation)CTR-GCN · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 84.9 -> 84.8No clear effect
use 5 CTR-GCs instead of 3 (ablation)CTR-GCN · NTU RGB+D 120 X-Sub (ablation)top-1 accuracy (%): 84.9 -> 85.4Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 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