Hi. I run MARS model on your current mmWave data to predict keypoint. MAE loss I calculated is around 15cm and MPJPE loss is around 1000mm. But in your paper, the MPJPE is just around 125-186mm.
The MPJPE formula I use is on image above.
Do I use wrong formula?
And how could I load your pretrained mmWave model?
Hi. I run MARS model on your current mmWave data to predict keypoint. MAE loss I calculated is around 15cm and MPJPE loss is around 1000mm. But in your paper, the MPJPE is just around 125-186mm.
The MPJPE formula I use is on image above.
Do I use wrong formula?
And how could I load your pretrained mmWave model?