search

kaichen-z / VMLoc

[AAAI 2021] VMLoc: Variational Fusion For Learning-Based Multimodal Camera Localization
https://arxiv.org/pdf/2003.07289
41 stars 8 forks source link

Question about data processing #2

Closed LZL-CS closed 1 year ago

LZL-CS commented 1 year ago

Hi @kaichen-z, When you generate the image's corresponding depth image, do you use the ins.csv pose file or vo.csv pose file? Because the generated pointcloud values(x, y, z) are quite different when you choose different files. When choosing the ins.csv pose file, we get the pointcloud values(x, y, z) is relative coordinates. While choosing the vo.csv pose file, we get the pointcloud values(x, y, z) is absolute coordinates.

I appreciate any help you can provide!

https://github.com/samarth-robo/robotcar-dataset-sdk/blob/86b251284b5583182bcd9584952d5421cdfb0040/python/project_laser_into_camera.py#L29C24-L29C34

image

kaichen-z commented 1 year ago

Hi, @LZL-CS. Use vo.csv, please.

LZL-CS commented 1 year ago

Hi, @LZL-CS. Use vo.csv, please.

OK, thanks for your quick reply!