Table 4.
Comparison with the state of the art on the BoLD validation and test set
| Set | Method | Year | mAP (%) | mRA (%) |
|---|---|---|---|---|
| Validation | TSN29 | 2018 | 18.55 | 64.27 |
| Filntisis et al.8 | 2020 | 16.56 | 62.66 | |
| Pikoulis et al.38 | 2021 | 19.30 | 66.94 | |
| Beyan et al.20,b | 2021 | 15.86 | 62.63 | |
| MANet | 2023 | 21.25 | 68.32 | |
| Test | I3D39 | 2017 | 15.37 | 61.24 |
| TSN29 | 2018 | 17.02 | 62.70 | |
| ST-GCN40,b | 2018 | 12.63 | 55.96 | |
| Filntisis et al.8 | 2020 | 17.96 | 64.16 | |
| random forest6,b | 2020 | 13.59 | 57.71 | |
| Pikoulis et al.38,a | 2021 | 21.87 | 68.29 | |
| Beyan et al.20,b | 2021 | 16.73 | 62.17 | |
| MANet | 2023 | 22.11 | 67.69 | |
| MANeta | 2023 | 23.09 | 69.23 |
The numbers highlighted in bold represent the best performance in the validation or test set.
a
Represents model ensembles.
b
Represents skeleton-based method.