Hello,
Great work!
I tried reprojecting the XSens 3D joints into the images, and there is always an offset in the overlay.
What I did:
XSens Joint (Px in XSens World Wx)
↓ Align XSens world to Aria world
Joint in Aria World (W)
↓ Camera pose inverse
Joint in Camera Frame
↓ Fisheye projection
2D Pixel (u, v)
↓ Draw
Final Image with Overlay
Could this offset be due to error propagation in the hand-eye calibration? Since both Aria and XSens are in different world coordinates, and slight shift in the suit or the glasses could cause error accumulation in the transform.
Can the authors please confirm if this is expected? I would also like to know if the 2D projection was attempted by the authors and if they observed similar outcomes.
Looking forward to your reply. Thank you!
Hello,
Great work!
I tried reprojecting the XSens 3D joints into the images, and there is always an offset in the overlay.
What I did:
XSens Joint (Px in XSens World Wx)
↓ Align XSens world to Aria world
Joint in Aria World (W)
↓ Camera pose inverse
Joint in Camera Frame
↓ Fisheye projection
2D Pixel (u, v)
↓ Draw
Final Image with Overlay
Could this offset be due to error propagation in the hand-eye calibration? Since both Aria and XSens are in different world coordinates, and slight shift in the suit or the glasses could cause error accumulation in the transform.
Can the authors please confirm if this is expected? I would also like to know if the 2D projection was attempted by the authors and if they observed similar outcomes.
Looking forward to your reply. Thank you!