search

drsrinathsridhar / xnocs

Multiview Aggregation for Learning Category-Specific Shape Reconstruction, NeurIPS 2019
38 stars 7 forks source link

Question about the ground truth of camera pose #8

Closed taeyeopl closed 2 years ago

taeyeopl commented 2 years ago

Thanks for sharing good work. I have some questions related to your work.

Q1. Can you share your code of getting the ground truth of pose using NOCS and [1] ?? I saw you only mentioned using [1] in the supplementary material and I wonder how to implement details. I also wonder how many frames are used for getting the pose.

Q2.As you mentioned in future work, Is there any method evaluating camera pose related NOCS?? I would really appreciate it if you could share your work.

[1] Richard Hartley and Andrew Zisserman. Multiple view geometry in computer vision. Cambridge university press, 2003.

drsrinathsridhar commented 2 years ago

Hi there,

Here's some code that estimates camera pose from NOCS maps. That should answer both your questions.

https://github.com/drsrinathsridhar/tk3dv/blob/2076704fb07de062619432fce1bd3b3fe5b909cd/examples/visualizeNOCSMap.py#L115