Table 4 Comparisons on the COCO test-dev set.
From: An efficient and accurate 2D human pose estimation method using VTTransPose network
Method | Input size | #Params (M) | FLOPs (G) | AP | AP0.5 | AP0.75 | APM | APL |
|---|---|---|---|---|---|---|---|---|
G-RMI28 | 353 × 257 | 42.6 | 57 | 64.9 | 85.5 | 71.3 | 62.3 | 70.0 |
Integral29 | 256 × 256 | 45.0 | 11.0 | 67.8 | 88.2 | 74.8 | 63.9 | 74.0 |
CPN7 | 384 × 288 | 58.8 | 29.2 | 72.1 | 91.4 | 80.0 | 68.7 | 77.2 |
RMPE30 | 320 × 256 | 28.1 | 26.7 | 72.3 | 89.2 | 79.1 | 68.0 | 78.6 |
SimpleBaseline27 | 384 × 288 | 68.6 | 35.6 | 73.7 | 91.9 | 81.1 | 70.3 | 80.0 |
HRNet-W328 | 384 × 288 | 28.5 | 16.0 | 74.9 | 92.5 | 82.8 | 71.3 | 80.9 |
HRNet-W488 | 256 × 192 | 63.6 | 14.6 | 74.2 | 92.4 | 82.4 | 70.9 | 79.7 |
HRNet-W488 | 384 × 288 | 63.6 | 32.9 | 75.5 | 92.5 | 83.3 | 71.9 | 81.5 |
DarkPose31 | 384 × 288 | 63.6 | 32.9 | 76.2 | 92.5 | 83.6 | 72.5 | 82.4 |
TokenPose16 | 256 × 192 | 13.5 | 5.7 | 74.0 | 91.9 | 81.5 | 70.6 | 79.8 |
DistilPose-S18 | 256 × 192 | 5.4 | 2.38 | 71.0 | 91.0 | 78.9 | 67.5 | 76.8 |
DistilPose-L18 | 256 × 192 | 21.3 | 10.33 | 73.7 | 91.6 | 81.1 | 70.2 | 79.6 |
GTPose-B19 | 256 × 192 | 13.5 | – | 74.5 | 92.2 | 82.2 | 70.7 | 79.8 |
TransPose-H-A417 | 256 × 192 | 17.3 | 17.5 | 74.7 | 91.9 | 82.2 | 71.4 | 80.7 |
TransPose-H-A617 | 256 × 192 | 17.5 | 21.8 | 75.0 | 92.2 | 82.3 | 71.3 | 81.1 |
TransPose-H–S17 | 256 × 192 | 8.0 | 10.2 | 73.4 | 91.6 | 81.1 | 70.1 | 79.3 |
VTTranspose (ours) | 256 × 192 | 6.0 | 5.4 | 73.6 | 91.4 | 81.1 | 70.1 | 79.6 |
