search

ispc-lab / LiDAR4D

💫 [CVPR 2024] LiDAR4D: Dynamic Neural Fields for Novel Space-time View LiDAR Synthesis
https://dyfcalid.github.io/LiDAR4D
Apache License 2.0
142 stars 13 forks source link

question about dataset preprocess #5

Closed howthep closed 3 months ago

howthep commented 4 months ago

Hi,

I'm very interested in your work. I noticed that the code only uses the sequence 2013_05_28_drive_0000, and the frame splitting is hard-coded.

My questions are:

  1. Are the results in the paper evaluated on 2013_05_28_drive_0000 or all sequences?
  2. If they are evaluated on all sequences, how is the frame splitting done?

data/preprocess/generate_rangeview.py, line 73 image same file, line 112 image

dyfcalid commented 4 months ago

Due to the time-consuming reconstruction, we only selected representative scenes on sequence 2013_05_28_drive_0000 for experiments in the paper, where the static scenes are consistent with LiDAR-NeRF. However, the provided code can be easily extended to other scenes or other sequences, depending on your needs.