Back to feed
arXiv cs.LG
arXiv cs.LG
7/15/2026
Repairing Shape-Prior Shortcuts in Long-Range Single-Shot Fringe Projection Profilometry

Repairing Shape-Prior Shortcuts in Long-Range Single-Shot Fringe Projection Profilometry

Short summary

PhiCalNet fixes a shape-prior shortcut in single-shot fringe projection profilometry where networks recover depth from object boundaries instead of fringe phase. By outputting wrapped-phase representations mapped through a fixed differentiable calibration layer, it removes the shortcut architecturally rather than via loss penalties. PhiCalNet reduces object MAE 3.3x to 4.46 mm and introduces pixel-wise conformal uncertainty quantification for FPP, localizing error at wrap discontinuities.

  • •Architectural fix for shape-prior shortcut in fringe projection profilometry networks
  • •Outputs wrapped-phase representation through fixed differentiable calibration layer, reducing MAE 3.3x
  • •First pixel-wise conformal uncertainty quantification for FPP, cutting RMSE by 64% when rejecting top 5% pixels

Generated with AI, which can make mistakes.

#research-breakthrough
Read full article at arXiv cs.LG

Is this a good recommendation for you?

Comments

Failed to load comments. Please try again.

Explore more

Diagnosing and Repairing Shape-Prior Shortcuts in Long-Range Single-Shot Fringe Projection Profilometry
arXiv cs.LG
Operator-Aware Mixed-Precision Tolerance Calibration for Tensor Kernels
arXiv cs.LG
LiST: Lipschitz Scaling Training for Robust and Calibrated Neural Networks
arXiv cs.LG
Orthogonal Gradient Constraints Shape Noisy-Label Memorization Dynamics
arXiv cs.LG