search

tjiiv-cprg / EPro-PnP

[CVPR 2022 Oral, Best Student Paper] EPro-PnP: Generalized End-to-End Probabilistic Perspective-n-Points for Monocular Object Pose Estimation
https://www.youtube.com/watch?v=TonBodQ6EUU
Apache License 2.0
1.11k stars 106 forks source link

Some questions #64

Open WentangChen opened 1 year ago

WentangChen commented 1 year ago
  1. can I simply use EPro-PnP for other two-stage pose estimation network?
  2. if I wanna to do so , can I just code 'from ops.pnp.epropnp import EProPnP6DoF' and epropnp = EProPnP6DoF(...)? or what else should I do ?

Thank you very much, I am looking forward to your guidance.

Lakonik commented 1 year ago

Yes, you can plug EPro-PnP's solver into other two-stage network during inference (which may be faster for large batches) or use it to re-train the original network (which requires slightly modifying the original network for w2d prediction). You can refer to the demo code for the usage.