ML Research Wiki / Benchmarks / Action Recognition / NTU RGB+D

NTU RGB+D

Action Recognition Benchmark

Performance Over Time

📊 Showing 21 results | 📏 Metric: Accuracy (CS)

Top Performing Models

Rank	Model	Paper	Accuracy (CS)	Date	Code
1	PoseC3D (RGB + Pose)	Revisiting Skeleton-based Action Recognition	97.00	2021-04-28	📦 open-mmlab/mmaction2 📦 kennymckormick/pyskl 📦 txyugood/PaddlePoseC3D 📦 sandman002/One-Style-is-All-You-Need-to-Generate-a-Video
2	π-ViT (RGB + Pose) 📚	Just Add $π$! Pose Induced Video Transformers for Understanding Activities of Daily Living	96.30	2023-11-30	📦 dominickrei/pi-vit
3	UMDR (RGB-D)	A Unified Multimodal De- and Re-coupling Framework for RGB-D Motion Recognition	96.20	2022-11-16	📦 zhoubenjia/motionrgbd-pami
4	EPAM-Net	EPAM-Net: An Efficient Pose-driven Attention-guided Multimodal Network for Video Action Recognition	96.10	2024-08-10	📦 ahmed-nady/multimodal-action-recognition
5	Hierarchical Action Classification (RGB + Pose)	Hierarchical Action Classification with Network Pruning	95.66	2020-07-30	-
6	VPN (RGB + Pose) 📚	VPN: Learning Video-Pose Embedding for Activities of Daily Living	95.50	2020-07-06	📦 srijandas07/VPN
7	EPP-Net (Parsing + Pose)	Explore Human Parsing Modality for Action Recognition	94.70	2024-01-04	📦 liujf69/EPP-Net-Action
8	3DA (RGB + Pose)	Cross-Modal Learning with 3D Deformable Attention for Action Recognition	94.30	2022-12-12	-
9	Action Machine (RGB only)	Action Machine: Rethinking Action Recognition in Trimmed Videos	94.30	2018-12-14	-
10	π-ViT (RGB only) 📚	Just Add $π$! Pose Induced Video Transformers for Understanding Activities of Daily Living	94.00	2023-11-30	📦 dominickrei/pi-vit

All Papers (21)

Revisiting Skeleton-based Action Recognition

2021

PoseC3D (RGB + Pose)

open-mmlab/mmaction2 kennymckormick/pyskl

Just Add $π$! Pose Induced Video Transformers for Understanding Activities of Daily Living

2023

π-ViT (RGB + Pose)

dominickrei/pi-vit

A Unified Multimodal De- and Re-coupling Framework for RGB-D Motion Recognition

2022

UMDR (RGB-D)

zhoubenjia/motionrgbd-pami

EPAM-Net: An Efficient Pose-driven Attention-guided Multimodal Network for Video Action Recognition

2024

EPAM-Net

ahmed-nady/multimodal-action-recognition

Hierarchical Action Classification with Network Pruning

2020

Hierarchical Action Classification (RGB + Pose)

VPN: Learning Video-Pose Embedding for Activities of Daily Living

2020

VPN (RGB + Pose)

srijandas07/VPN

Explore Human Parsing Modality for Action Recognition

2024

EPP-Net (Parsing + Pose)

liujf69/EPP-Net-Action

Cross-Modal Learning with 3D Deformable Attention for Action Recognition

2022

3DA (RGB + Pose)

Action Machine: Rethinking Action Recognition in Trimmed Videos

2018

Action Machine (RGB only)

Just Add $π$! Pose Induced Video Transformers for Understanding Activities of Daily Living

2023

π-ViT (RGB only)

dominickrei/pi-vit

Integrating Human Parsing and Pose Network for Human Action Recognition

2023

IPP-Net (Parsing + Pose)

liujf69/ipp-net-parsing

DVANet: Disentangling View and Action Features for Multi-View Action Recognition

2023

DVANet (RGB only)

NyleSiddiqui/MultiView_Actions

DSTSA-GCN: Advancing Skeleton-Based Gesture Recognition with Semantic-Aware Spatio-Temporal Topology Modeling

2025

DSTSA-GCN

HuCui2022/DSTSA-GCN_Gesture

MSAF: Multimodal Split Attention Fusion

2020

MSAF (RGB+Pose)

anita-hu/MSAF

STAR-Transformer: A Spatio-temporal Cross Attention Transformer for Human Action Recognition

2022

STAR-Transformer (RGB + Pose)

MMTM: Multimodal Transfer Module for CNN Fusion

2019

MMTM (RGB+Pose)

haamoon/mmtm

Infrared and 3D skeleton feature fusion for RGB-D action recognition

2020

FUSION (IR+Pose)

adeboissiere/FUSION-human-action-recognition

Part-based Graph Convolutional Network for Action Recognition

2018

PB-GCN (Skeleton only)

kalpitthakkar/pb-gcn

Glimpse Clouds: Human Activity Recognition from Unstructured Feature Points

2018

Glimpse Clouds (RGB only)

fabienbaradel/glimpse_clouds

SkeleMotion: A New Representation of Skeleton Joint Sequences Based on Motion Information for 3D Action Recognition

2019

Skelemotion + Yang et al. (Skeleton only)

carloscaetano/skeleton-images

Deep Multimodal Feature Analysis for Action Recognition in RGB+D Videos

2016

DSSCA-SSLM (RGB only)

NTU RGB+D

Performance Over Time

Edit Benchmark Results

Edit Result

Top Performing Models

All Papers (21)

Revisiting Skeleton-based Action Recognition

Just Add $π$! Pose Induced Video Transformers for Understanding Activities of Daily Living

A Unified Multimodal De- and Re-coupling Framework for RGB-D Motion Recognition

EPAM-Net: An Efficient Pose-driven Attention-guided Multimodal Network for Video Action Recognition

Hierarchical Action Classification with Network Pruning

VPN: Learning Video-Pose Embedding for Activities of Daily Living

Explore Human Parsing Modality for Action Recognition

Cross-Modal Learning with 3D Deformable Attention for Action Recognition

Action Machine: Rethinking Action Recognition in Trimmed Videos

Just Add $π$! Pose Induced Video Transformers for Understanding Activities of Daily Living

Integrating Human Parsing and Pose Network for Human Action Recognition

DVANet: Disentangling View and Action Features for Multi-View Action Recognition

DSTSA-GCN: Advancing Skeleton-Based Gesture Recognition with Semantic-Aware Spatio-Temporal Topology Modeling

MSAF: Multimodal Split Attention Fusion

STAR-Transformer: A Spatio-temporal Cross Attention Transformer for Human Action Recognition

MMTM: Multimodal Transfer Module for CNN Fusion

Infrared and 3D skeleton feature fusion for RGB-D action recognition

Part-based Graph Convolutional Network for Action Recognition

Glimpse Clouds: Human Activity Recognition from Unstructured Feature Points

SkeleMotion: A New Representation of Skeleton Joint Sequences Based on Motion Information for 3D Action Recognition

Deep Multimodal Feature Analysis for Action Recognition in RGB+D Videos

Model	Paper	Accuracy (CS)	Date
PoseC3D (RGB + Pose)	Revisiting Skeleton-based Action Recognition	97.00	2021-04-28
π-ViT (RGB + Pose)	Just Add $π$! Pose Induced Video Transformers for…	96.30	2023-11-30
UMDR (RGB-D)	A Unified Multimodal De- and Re-coupling Framewor…	96.20	2022-11-16
EPAM-Net	EPAM-Net: An Efficient Pose-driven Attention-guid…	96.10	2024-08-10
Hierarchical Action Classification (RGB + Pose)	Hierarchical Action Classification with Network P…	95.66	2020-07-30
VPN (RGB + Pose)	VPN: Learning Video-Pose Embedding for Activities…	95.50	2020-07-06
EPP-Net (Parsing + Pose)	Explore Human Parsing Modality for Action Recogni…	94.70	2024-01-04
3DA (RGB + Pose)	Cross-Modal Learning with 3D Deformable Attention…	94.30	2022-12-12
Action Machine (RGB only)	Action Machine: Rethinking Action Recognition in …	94.30	2018-12-14
π-ViT (RGB only)	Just Add $π$! Pose Induced Video Transformers for…	94.00	2023-11-30
IPP-Net (Parsing + Pose)	Integrating Human Parsing and Pose Network for Hu…	93.80	2023-07-16
DVANet (RGB only)	DVANet: Disentangling View and Action Features fo…	93.40	2023-12-10
DSTSA-GCN	DSTSA-GCN: Advancing Skeleton-Based Gesture Recog…	92.78	2025-01-21
MSAF (RGB+Pose)	MSAF: Multimodal Split Attention Fusion	92.24	2020-12-13
STAR-Transformer (RGB + Pose)	STAR-Transformer: A Spatio-temporal Cross Attenti…	92.00	2022-10-14
MMTM (RGB+Pose)	MMTM: Multimodal Transfer Module for CNN Fusion	91.99	2019-11-20
FUSION (IR+Pose)	Infrared and 3D skeleton feature fusion for RGB-D…	91.80	2020-02-28
PB-GCN (Skeleton only)	Part-based Graph Convolutional Network for Action…	87.50	2018-09-13
Glimpse Clouds (RGB only)	Glimpse Clouds: Human Activity Recognition from U…	86.60	2018-02-22
Skelemotion + Yang et al. (Skeleton only)	SkeleMotion: A New Representation of Skeleton Joi…	76.50	2019-07-30
DSSCA-SSLM (RGB only)	Deep Multimodal Feature Analysis for Action Recog…	74.90	2016-03-23