Table  1.

The performance of the latest state-of-the-art 3D skeleton-based methods on NTU-RGB+D dataset

NTU-RGB+D dataset
Rank	Paper	Year	Accuracy (C-View)	Accuracy (C-Subject)	Method
1	Wang et al. [55]	2023	98.7	94.8	Two-stream Transformer
2	Duan et al. [134]	2022	97.5	93.2	Dynamic group GCN
3	Liu et al. [135]	2023	96.8	92.8	Temporal decoupling GCN
4	Zhou et al. [56]	2022	96.5	92.9	Transformer
5	Chen et al. [136]	2021	96.8	92.4	Topology refinement GCN
6	Zeng et al. [137]	2021	96.7	91.6	Skeletal GCN
7	Liu et al. [103]	2020	96.2	91.5	Disentangling and unifying GCN
8	Ye et al. [138]	2020	96.0	91.5	Dynamic GCN
9	Shi et al. [139]	2019	96.1	89.9	Directed graph neural networks
10	Shi et al. [95]	2018	95.1	88.5	Two-stream adaptive GCN
11	Zhang et al. [140]	2018	95.0	89.2	LSTM-based RNN
12	Si et al. [141]	2019	95.0	89.2	AGC-LSTM(Joints&Part)
13	Hu et al. [142]	2018	94.9	89.1	Nonlocal S-T + frequency attention
14	Li et al. [101]	2019	94.2	86.8	GCN
15	Liang et al. [143]	2019	93.7	88.6	3S-CNN + multitask ensemble learning
16	Song et al. [144]	2019	93.5	85.9	Richly activated GCN
17	Zhang et al. [145]	2019	93.4	86.6	Semantics-guided GCN
18	Xie et al. [77]	2018	93.2	82.7	RNN+CNN+Attention