Meet a problem when running Nuscenes data preparation script.

See original GitHub issue

Environment:

Python = 3.8
PyTorch = 1.10.1
tqdm = 4.64.0
torchpack = 0.3.1
mmcv-full = 1.4.0
mmdet = 2.20.0
mmdet3d = 0.0.0
torchvision = 0.11.2

Description: After I prepare the develop environment and download the Nuscenes original dataset, I try to run the command:python tools/create_data.py nuscenes --root-path ./data/nuscenes --out-dir ./data/nuscenes --extra-tag nuscenes, but I meet an error :Traceback (most recent call last): File "tools/create_data.py", line 3, in <module> from tools.data_converter import nuscenes_converter as nuscenes_converter File "/home/user/BEVFusion/tools/data_converter/nuscenes_converter.py", line 13, in <module> from mmdet3d.core.bbox.box_np_ops import points_cam2img File "/home/user/BEVFusion/mmdet3d/core/__init__.py", line 2, in <module> from .bbox import * # noqa: F401, F403 File "/home/user/BEVFusion/mmdet3d/core/bbox/__init__.py", line 3, in <module> from .iou_calculators import (AxisAlignedBboxOverlaps3D, BboxOverlaps3D, File "/home/user/BEVFusion/mmdet3d/core/bbox/iou_calculators/__init__.py", line 1, in <module> from .iou3d_calculator import ( File "/home/user/BEVFusion/mmdet3d/core/bbox/iou_calculators/iou3d_calculator.py ", line 5, in <module> from ..structures import get_box_type File "/home/user/BEVFusion/mmdet3d/core/bbox/structures/__init__.py", line 1, in <module> from .base_box3d import BaseInstance3DBoxes File "/home/user/BEVFusion/mmdet3d/core/bbox/structures/base_box3d.py", line 5, in <module> from mmdet3d.ops.iou3d import iou3d_cuda File "/home/user/BEVFusion/mmdet3d/ops/__init__.py", line 34, in <module> from .roiaware_pool3d import ( File "/home/user/BEVFusion/mmdet3d/ops/roiaware_pool3d/__init__.py", line 1, in <module> from .points_in_boxes import points_in_boxes_batch, points_in_boxes_cpu, points_in_boxes_gpu File "/home/user/BEVFusion/mmdet3d/ops/roiaware_pool3d/points_in_boxes.py", line 3, in <module> from . import roiaware_pool3d_ext ImportError: /home/user/BEVFusion/mmdet3d/ops/roiaware_pool3d/roiaware_pool3d_ext. cpython-38-x86_64-linux-gnu.so: undefined symbol: _ZN3c1015SmallVectorBaseIjE8grow_podEPvmm . Many thanks for your kindly help!

Issue Analytics

  • State:closed
  • Created a year ago
  • Comments:8 (4 by maintainers)

github_iconTop GitHub Comments

1reaction
PigLogic-Cybercommented, Jun 13, 2022

Hi @PigLogic-Cyber,

Have you tried deleting the build folder and recompile? It seems to me that the setup.py script did nothing but copied pre-built results around. However, as you have mentioned, you changed the version of the CUDA toolkit. In this case, I believe a re-compilation is necessary.

Related to the library version, actually I’m also using PyTorch 1.10.1 + CUDA devkit 11.3. So I don’t think your library version is problematic.

If thing still cannot work, I would suggest you to try compiling other mmdetection3d-based repos, such as DETR3D. I’m curious whether similar error would occur again when you

from mmdet3d.ops.iou3d import iou3d_cuda

Best, Haotian

Dear Haotian, After deleting the build dir and recompile the development environment, problem solved. So my original problem was caused by the higher version of pytorch(1.11.0). Then the problem happended since I downgraded the pytorch version but forgot to delete build dir before compiling the development environment. Really save my day, thx a lot! Best, Roy

0reactions
kentang-mitcommented, Jun 13, 2022

Hi @PigLogic-Cyber,

Have you tried deleting the build folder and recompile? It seems to me that the setup.py script did nothing but copied pre-built results around. However, as you have mentioned, you changed the version of the CUDA toolkit. In this case, I believe a re-compilation is necessary.

Related to the library version, actually I’m also using PyTorch 1.10.1 + CUDA devkit 11.3. So I don’t think your library version is problematic.

If thing still cannot work, I would suggest you to try compiling other mmdetection3d-based repos, such as DETR3D. I’m curious whether similar error would occur again when you

from mmdet3d.ops.iou3d import iou3d_cuda

Best, Haotian

Read more comments on GitHub >

github_iconTop Results From Across the Web

AB3DMOT/nuScenes.md at master - GitHub
3D Multi-Object Tracking. Once the data is properly prepared, everything is as easy as running for KITTI inference. For example, one can run...
Read more >
nuScenes prediction task
The goal of the nuScenes prediction task is to predict the future trajectories of objects in the nuScenes dataset. A trajectory is a...
Read more >
v1.0.0rc4 — MMDetection3D 1.0.0rc4 documentation
We have adopted CUDA operators compiled from mmcv and removed all the CUDA operators in ... Please re-run the nuScenes data preparation script...
Read more >
mmdet Changelog - pyup.io
The modification only influences nuScenes coco-style json files. Please re-run the data preparation scripts if necessary. See more details in the PR 744....
Read more >
How to install this Python package from sources?
TL;DR: Create a setup.py file in the root directory of the nuscenes-devkit repository with the following content (then, to install, run pip ...
Read more >

github_iconTop Related Medium Post

No results found

github_iconTop Related StackOverflow Question

No results found

github_iconTroubleshoot Live Code

Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start Free

github_iconTop Related Reddit Thread

No results found

github_iconTop Related Hackernoon Post

No results found

github_iconTop Related Tweet

No results found

github_iconTop Related Dev.to Post

No results found

github_iconTop Related Hashnode Post

No results found