search

QianyiWu / objsdf

:t-rex: [ECCV‘22] Pytorch implementation of 'Object-Compositional Neural Implicit Surfaces'
183 stars 5 forks source link

Question about center.txt #5

Closed SuzyZhao0417 closed 1 year ago

SuzyZhao0417 commented 1 year ago

Hi, Thanks for your amazing work! I'm curious about the content in your center.txt. From your code, the content is related to scale_mat and pose_matrix. As my own understanding it can help generate the sample sphere, could you please describe it clearly? Thank you for your reply, best wishes!

QianyiWu commented 1 year ago

Hi, the first three elements in 'center txt' is the coordinate of the center point in the original scene. The last element is the size of the whole scene. After the transformation, the entire scene will be located inside a sphere with a radius of 3.

This preprocessing is very similar to here and here. The main purpose is to make the whole scene in a suitable range.

SuzyZhao0417 commented 1 year ago

Hi, the first three elements in 'center txt' is the coordinate of the center point in the original scene. The last element is the size of the whole scene. After the transformation, the entire scene will be located inside a sphere with a radius of 3.

This preprocessing is very similar to here and here. The main purpose is to make the whole scene in a suitable range.

Thanks for your reply! I understand the meaning

rockywind commented 1 year ago

Hi, the first three elements in 'center txt' is the coordinate of the center point in the original scene. The last element is the size of the whole scene. After the transformation, the entire scene will be located inside a sphere with a radius of 3.

This preprocessing is very similar to here and here. The main purpose is to make the whole scene in a suitable range.

Hi, how to modified the sphere's radius? I wanted to training the own dataset. 0001

QianyiWu commented 1 year ago

Hi, the first three elements in 'center txt' is the coordinate of the center point in the original scene. The last element is the size of the whole scene. After the transformation, the entire scene will be located inside a sphere with a radius of 3. This preprocessing is very similar to here and here. The main purpose is to make the whole scene in a suitable range.

Hi, how to modified the sphere's radius? I wanted to training the own dataset. 0001

You can modify the sphere's radius in the config file. But a better choice should be to scale the scene coordinate system to make it stay inside a sphere with a radius of 3.

But currently, our method can not support getting object SDF from only a single image. It requires multiview images with camera pose and the corresponding instance mask for training.