Mis-classified log files for sem2lidar and cam2lidar in README.md

[liuQuan98]

Great work and open-source! I really appreciate it.

However, I found that your produced log files for sem2lidar and cam2lidar are swapped in the README file, i.e., the 5.4 Gig log contains sem2lidar (because the original_conditioning.pngs are semantic maps) and the 2.1 Gig log contains cam2lidar, but they are listed as the exact opposite in the table. Could you please update the README.md to fix it? I also want to try if the model ckpts are swapped as well, but sadly I have no gpus at hand and cannot test it. I would appreciate it if you could be kind enough to check that as well.

[Haiyunn]

@liuQuan98 Hello, have you resolved the issue mentioned above? I found that in the cam2lidar log file, there is a mismatch between the images and the sampled point clouds, and they do not correspond one-to-one. I am not sure whether this is due to the author incorrectly labeling the frames or if it is caused by inaccurate cam2lidar sampling.

[liuQuan98]

@liuQuan98 Hello, have you resolved the issue mentioned above? I found that in the cam2lidar log file, there is a mismatch between the images and the sampled point clouds, and they do not correspond one-to-one. I am not sure whether this is due to the author incorrectly labeling the frames or if it is caused by inaccurate cam2lidar sampling.

@Haiyunn No, as I no longer use the official logs now. It's easy to train and generate data with our own gpus.

[Haiyunn]

@liuQuan98 Thank you for your reply. I’m wondering if it would be convenient for you to send the results of your sampling, as I would like to compare them. Due to time and GPU limitations, I haven’t been able to reproduce this work yet. Thank you very much