Skip to main content

Table 3 PCK performance comparison on the validation and test dataset. The table shows the thresholds of 0.1, 0.2, and 0.3. Higher value is better in this metrics

From: 3D human pose estimation model using location-maps for distorted and disconnected images by a wearable omnidirectional camera

Model Backbone PCK @ 0.1 PCK @ 0.2 PCK @ 0.3
Validation dataset
VNect ResNet50 91.98 98.53 99.56
VNect ResNet100 92.42 98.64 99.60
Ours HRNet-W24 94.02 98.79 99.60
Ours HRNet-W32 93.91 98.75 99.59
Test dataset
VNect ResNet50 48.11 65.74 77.63
VNect ResNet100 49.49 67.01 78.11
Ours HRNet-W24 49.65 67.07 77.89
Ours HRNet-W32 50.45 68.09 79.12
  1. The best performance estimations are italicized