search

lyft / nuscenes-devkit

Devkit for the public 2019 Lyft Level 5 AV Dataset (fork of https://github.com/nutonomy/nuscenes-devkit)
Other
375 stars 101 forks source link

Incorrect map path on kaggle dataset #17

Open stalkermustang opened 5 years ago

stalkermustang commented 5 years ago

Hello. image image Here we see that map path in train json is incorrect. Also in kernels i see this code:

!ln -s $DATASET_PATH/train_images images
!ln -s $DATASET_PATH/train_maps maps
!ln -s $DATASET_PATH/train_lidar lidar

Maybe paths other than train_maps also broken

vgoklani commented 5 years ago

same issue here

vgoklani commented 5 years ago

The issue is line 85: 

map_record["mask"] = MapMask(self.data_path / map_record["filename"], resolution=map_resolution)

I suspect that maps was incorrectly mapped to train/test_maps

dekuska10 commented 5 years ago

@vgoklani @stalkermustang i am getting this error "map mask /maps/map_raster_palo_alto.png does not exist" have you guys solved this

stalkermustang commented 5 years ago

link folder with map into /map/ dir

ixtiyoruz commented 4 years ago

after linking, change the loading like this: lyft_dataset = LyftDataset(data_path='./', json_path=DATA_PATH +'train_data', verbose=True)

ruhyadi commented 2 years ago

Have someone solve this?. If not, I provide my solution. Please symlink kaggle dataset to working directory. After that, use working directory as input path

# symlink 
DATASET_PATH = '/kaggle/input/3d-object-detection-for-autonomous-vehicles'
DEST_PATH = '/kaggle/working/'

!ln -s $DATASET_PATH/train_images $DEST_PATH/images
!ln -s $DATASET_PATH/train_data $DEST_PATH/data
!ln -s $DATASET_PATH/train_lidar $DEST_PATH/lidar
!ln -s $DATASET_PATH/train_maps $DEST_PATH/maps
# load lyft sdk
from lyft_dataset_sdk.lyftdataset import LyftDataset

# load dataset
level5data = LyftDataset(data_path='./', json_path='./data', verbose=True)