Closed LiuXuan66 closed 6 months ago
Hi, we selected static scenes in the Waymo dataset as shown in process_waymo_script
.
You just need to download some of them and then continue.
Hi, we selected static scenes in the Waymo dataset as shown in
process_waymo_script
.You just need to download some of them and then continue.
So could you please offer the paths of these scenes in waymo google cloud, I can't find some scenes in training or val set. And are you sure the data is in waymo v1.4.2 individual files not in other version?
Sorry for not clearly explaining how to download the data. Yes, we use Waymo v1.4.2 individual files.
'segment-10247954040621004675_2180_000_2200_000_with_camera_labels.tfrecord' 'segment-13469905891836363794_4429_660_4449_660_with_camera_labels.tfrecord' 'segment-14333744981238305769_5658_260_5678_260_with_camera_labels.tfrecord'
These three tfrecord files are from the validation set. And the rest of the list are from the training set.
Maybe you can try to use gsutil
to download the data like this:
gsutil cp -n gs://waymo_open_dataset_v_1_4_2/individual_files/validation/segment-10247954040621004675_2180_000_2200_000_with_camera_labels.tfrecord ${your_dest_file_path}
In addition, if you want to download the files from their website directly, you can use the filter
function to search the corresponding tfrecord file names and download them.
Thanks for you reply. I'm new to Waymo, and I thought there were 50 files in training because only 50 files are shown in the first page, and now I see the bottom of page that there are 798 files.
Yes, the whole Waymo dataset includes about 1000 sequences (train + validation). Most of them have dynamic objects and we only select those static scenes.
Got it! Thank you very much!
below is process_waymo_script: I download the waymo dataset as your readme:v1.4.2/individual_files/training, but they are different with the scene names in process_waymo_scrip.py, so I can't execute the command to get correct data.