Papers › Revisiting Skeleton-based Action Recognition

Revisiting Skeleton-based Action Recognition

28 Apr 2021CVPR 2022 1arXiv:2104.13586archive 2025-07-28

Haodong Duan, Yue Zhao, Kai Chen, Dahua Lin, Bo Dai

Human skeleton, as a compact representation of human action, has received increasing attention in recent years. Many skeleton-based action recognition methods adopt graph convolutional networks (GCN) to extract features on top of human skeletons. Despite the positive results shown in previous works, GCN-based methods are subject to limitations in robustness, interoperability, and scalability. In this work, we propose PoseC3D, a new approach to skeleton-based action recognition, which relies on a 3D heatmap stack instead of a graph sequence as the base representation of human skeletons. Compared to GCN-based methods, PoseC3D is more effective in learning spatiotemporal features, more robust against pose estimation noises, and generalizes better in cross-dataset settings. Also, PoseC3D can handle multiple-person scenarios without additional computation cost, and its features can be easily integrated with other modalities at early fusion stages, which provides a great design space to further boost the performance. On four challenging datasets, PoseC3D consistently obtains superior performance, when used alone on skeletons and in combination with the RGB modality.

PaperPDFConference PDFCode

In Syntology View this paper on Syntology, its page in Syntology's graph. That page lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

kennymckormick/pyskl mentioned in papermentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Channel-wise Topology Refinement Graph Convolution for Skeleton-Based Action Recognition (arXiv:2107.12213; that paper's own run record: 2 ran (of which 2 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 0 where Syntology's instrument failed) · 3 unverified; every one of the 2 samples that ran constructed an object rather than computing a result); nothing checks that they implement this paper's method. report
open-mmlab/mmaction2 mentioned in paperpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Temporal Interlacing Network (arXiv:2001.06499; that paper's own run record: 17 ran (of which 9 constructed an object rather than computing a result; 14 with no instrument failure: 0 honoured, 0 violated, 14 with no contract checked; 3 where Syntology's instrument failed) · 8 unverified); nothing checks that they implement this paper's method. report
sandman002/One-Style-is-All-You-Need-to-Generate-a-Video mentioned on GitHubpytorchSyntology: not harvested report
txyugood/PaddlePoseC3D paddleSyntology: not harvested report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

3D Action RecognitionAction RecognitionGroup Activity RecognitionPose EstimationSkeleton Based Action Recognition

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
3D Action Recognition Assembly101 RGBPoseConv3D Actions Top-1 33.61 #3 of 7 Archive leaderboard report
3D Action Recognition Assembly101 RGBPoseConv3D Object Top-1 42.90 #3 of 7 Archive leaderboard report
3D Action Recognition Assembly101 RGBPoseConv3D Verbs Top-1 61.99 #3 of 7 Archive leaderboard report
Action Recognition H2O (2 Hands and Objects) RGBPoseConv3D Actions Top-1 83.47 #7 of 11 Archive leaderboard report
Action Recognition H2O (2 Hands and Objects) RGBPoseConv3D Hand Pose 2D #7 of 11 Archive leaderboard report
Action Recognition H2O (2 Hands and Objects) RGBPoseConv3D Object Label No #7 of 11 Archive leaderboard report
Action Recognition H2O (2 Hands and Objects) RGBPoseConv3D Object Pose No #7 of 11 Archive leaderboard report
Action Recognition H2O (2 Hands and Objects) RGBPoseConv3D RGB Yes #7 of 11 Archive leaderboard report
Action Recognition NTU RGB+D PoseC3D (RGB + Pose) Accuracy (CS) 97.0 #2 of 28 Archive leaderboard report
Action Recognition NTU RGB+D PoseC3D (RGB + Pose) Accuracy (CV) 99.6 #2 of 28 Archive leaderboard report
Action Recognition NTU RGB+D 120 PoseC3D (RGB + Pose) Accuracy (Cross-Setup) 96.4 #2 of 21 Archive leaderboard report
Action Recognition NTU RGB+D 120 PoseC3D (RGB + Pose) Accuracy (Cross-Subject) 95.3 #2 of 21 Archive leaderboard report
Action Recognition Volleyball PoseC3D (Pose Only) Accuracy 91.3 #1 of 4 Archive leaderboard report
Group Activity Recognition Volleyball PoseC3D (Pose-Only) Accuracy 91.3 #8 of 12 Archive leaderboard report
Skeleton Based Action Recognition Kinetics-Skeleton dataset PoseC3D (SlowOnly-346) Accuracy 49.1 #4 of 42 Archive leaderboard report
Skeleton Based Action Recognition Kinetics-Skeleton dataset PoseC3D Accuracy 47.7 #5 of 42 Archive leaderboard report
Skeleton Based Action Recognition NTU RGB+D PoseC3D [3D Heatmap] Accuracy (CS) 94.1 #2 of 135 Archive leaderboard report
Skeleton Based Action Recognition NTU RGB+D PoseC3D [3D Heatmap] Accuracy (CV) 97.1 #2 of 135 Archive leaderboard report
Skeleton Based Action Recognition NTU RGB+D PoseC3D [3D Heatmap] Ensembled Modalities 2 #2 of 135 Archive leaderboard report
Skeleton Based Action Recognition NTU RGB+D 120 PoseC3D (w. HRNet 2D Skeleton) Accuracy (Cross-Setup) 90.3 #37 of 83 Archive leaderboard report
Skeleton Based Action Recognition NTU RGB+D 120 PoseC3D (w. HRNet 2D Skeleton) Accuracy (Cross-Subject) 86.9 #37 of 83 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Graph Convolutional NetworksHeatmap

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections