Table 8 Comparison of our DSwarm-Net model with some existing methods on the NTU RGB+D 60 dataset.
From: A union of deep learning and swarm-based optimization for 3D human action recognition
Method | Year | Cross-subject accuracy (%) | Cross-view accuracy (%) |
|---|---|---|---|
STVA LSTM23 | 2019 | 82.40 | 89.10 |
Deep STGC27 | 2019 | 86.45 | 84.65 |
PC Net75 | 2019 | 85.25 | 91.37 |
Shift GCN32 | 2020 | 90.70 | 96.5 |
DS LSTM33 | 2020 | 77.79 | 87.44 |
AGC-LSTM76 | 2020 | 89.20 | 95.00 |
PA-ResGCN-B1977 | 2020 | 90.90 | 96.00 |
MV-IGNet78 | 2020 | 89.2 | 96.3 |
VIDA79 | 2020 | 79.40 | 84.10 |
MS-G3D31 | 2020 | 91.5 | 96.2 |
CTR-GCN80 | 2021 | 92.4 | 96.8 |
EfficientGCN-B481 | 2021 | 91.7 | 95.7 |
ST-TR82 | 2021 | 89.91 | 93.1 |
DSwarm-Net | 2021 | 85.45 | 89.98 |
