Cannot import name 'ball_query_ext' with official Docker environment

[Dinghow]

Hi,

I successfully built the environment from your official Dockfile

ARG PYTORCH="1.6.0"
ARG CUDA="10.1"
ARG CUDNN="7"

FROM pytorch/pytorch:${PYTORCH}-cuda${CUDA}-cudnn${CUDNN}-devel

ENV TORCH_CUDA_ARCH_LIST="6.0 6.1 7.0+PTX"
ENV TORCH_NVCC_FLAGS="-Xfatbin -compress-all"
ENV CMAKE_PREFIX_PATH="$(dirname $(which conda))/../"

RUN apt-get update && apt-get install -y ffmpeg libsm6 libxext6 git ninja-build libglib2.0-0 libsm6 libxrender-dev libxext6 \
    && apt-get clean \
    && rm -rf /var/lib/apt/lists/*

# Install MMCV
RUN pip install mmcv-full==latest+torch1.6.0+cu101 -f https://openmmlab.oss-accelerate.aliyuncs.com/mmcv/dist/index.html
RUN pip install mmdet

# Install MMDetection
RUN conda clean --all
RUN git clone https://github.com/open-mmlab/mmdetection3d.git /mmdetection3d
WORKDIR /mmdetection3d
ENV FORCE_CUDA="1"
RUN pip install -r requirements/build.txt
RUN pip install --no-cache-dir -e .

# uninstall pycocotools installed by nuscenes-devkit and reinstall mmpycocotools
RUN pip uninstall pycocotools --no-cache-dir -y
RUN pip install mmpycocotools --no-cache-dir --force --no-deps
RUN pip install tensorboard

And I can successfully execute inference with your pre-trained model. However, when I try training with instruction ./tools/dist_train.sh configs/pointpillars/hv_pointpillars_secfpn_6x8_160e_kitti-3d-car.py 4, there will raise errors:

Traceback (most recent call last):
4/28/2021 10:06:37 PM File "./tools/train.py", line 16, in
4/28/2021 10:06:37 PM from mmdet3d.datasets import build_dataset
4/28/2021 10:06:37 PM File "/output/mmdet3d/datasets/init.py", line 3, in
4/28/2021 10:06:37 PM from .custom_3d import Custom3DDataset
4/28/2021 10:06:37 PM File "/output/mmdet3d/datasets/custom_3d.py", line 9, in
4/28/2021 10:06:37 PM from ..core.bbox import get_box_type
4/28/2021 10:06:37 PM File "/output/mmdet3d/core/init.py", line 2, in
4/28/2021 10:06:37 PM from .bbox import * # noqa: F401, F403
4/28/2021 10:06:37 PM File "/output/mmdet3d/core/bbox/init.py", line 4, in
4/28/2021 10:06:37 PM from .iou_calculators import (AxisAlignedBboxOverlaps3D, BboxOverlaps3D,
4/28/2021 10:06:37 PM File "/output/mmdet3d/core/bbox/iou_calculators/init.py", line 1, in
4/28/2021 10:06:37 PM from .iou3d_calculator import (AxisAlignedBboxOverlaps3D, BboxOverlaps3D,
4/28/2021 10:06:37 PM File "/output/mmdet3d/core/bbox/iou_calculators/iou3d_calculator.py", line 5, in
4/28/2021 10:06:37 PM from ..structures import get_box_type
4/28/2021 10:06:37 PM File "/output/mmdet3d/core/bbox/structures/init.py", line 1, in
4/28/2021 10:06:37 PM from .base_box3d import BaseInstance3DBoxes
4/28/2021 10:06:37 PM File "/output/mmdet3d/core/bbox/structures/base_box3d.py", line 5, in
4/28/2021 10:06:37 PM from mmdet3d.ops.iou3d import iou3d_cuda
4/28/2021 10:06:37 PM File "/output/mmdet3d/ops/init.py", line 5, in
4/28/2021 10:06:37 PM from .ball_query import ball_query
4/28/2021 10:06:37 PM File "/output/mmdet3d/ops/ball_query/init.py", line 1, in
4/28/2021 10:06:37 PM from .ball_query import ball_query
4/28/2021 10:06:37 PM File "/output/mmdet3d/ops/ball_query/ball_query.py", line 4, in
4/28/2021 10:06:37 PM from . import ball_query_ext
4/28/2021 10:06:37 PM ImportError: cannot import name 'ball_query_ext' from 'mmdet3d.ops.ball_query' (/output/mmdet3d/ops/ball_query/init.py)
4/28/2021 10:06:38 PM Traceback (most recent call last):
4/28/2021 10:06:38 PM File "./tools/train.py", line 16, in
4/28/2021 10:06:38 PM from mmdet3d.datasets import build_dataset
4/28/2021 10:06:38 PM File "/output/mmdet3d/datasets/init.py", line 3, in
4/28/2021 10:06:38 PM from .custom_3d import Custom3DDataset
4/28/2021 10:06:38 PM File "/output/mmdet3d/datasets/custom_3d.py", line 9, in
4/28/2021 10:06:38 PM from ..core.bbox import get_box_type
4/28/2021 10:06:38 PM File "/output/mmdet3d/core/init.py", line 2, in
4/28/2021 10:06:38 PM from .bbox import * # noqa: F401, F403
4/28/2021 10:06:38 PM File "/output/mmdet3d/core/bbox/init.py", line 4, in
4/28/2021 10:06:38 PM from .iou_calculators import (AxisAlignedBboxOverlaps3D, BboxOverlaps3D,
4/28/2021 10:06:38 PM File "/output/mmdet3d/core/bbox/iou_calculators/init.py", line 1, in
4/28/2021 10:06:38 PM from .iou3d_calculator import (AxisAlignedBboxOverlaps3D, BboxOverlaps3D,
4/28/2021 10:06:38 PM File "/output/mmdet3d/core/bbox/iou_calculators/iou3d_calculator.py", line 5, in
4/28/2021 10:06:38 PM from ..structures import get_box_type
4/28/2021 10:06:38 PM File "/output/mmdet3d/core/bbox/structures/init.py", line 1, in
4/28/2021 10:06:38 PM from .base_box3d import BaseInstance3DBoxes
4/28/2021 10:06:38 PM File "/output/mmdet3d/core/bbox/structures/base_box3d.py", line 5, in
4/28/2021 10:06:38 PM from mmdet3d.ops.iou3d import iou3d_cuda
4/28/2021 10:06:38 PM File "/output/mmdet3d/ops/init.py", line 5, in
4/28/2021 10:06:38 PM from .ball_query import ball_query
4/28/2021 10:06:38 PM File "/output/mmdet3d/ops/ball_query/init.py", line 1, in
4/28/2021 10:06:38 PM from .ball_query import ball_query
4/28/2021 10:06:38 PM File "/output/mmdet3d/ops/ball_query/ball_query.py", line 4, in
4/28/2021 10:06:38 PM from . import ball_query_ext
4/28/2021 10:06:38 PM ImportError: cannot import name 'ball_query_ext' from 'mmdet3d.ops.ball_query' (/output/mmdet3d/ops/ball_query/init.py)
4/28/2021 10:06:38 PM Traceback (most recent call last):
4/28/2021 10:06:38 PM File "/opt/conda/lib/python3.7/runpy.py", line 193, in _run_module_as_main
4/28/2021 10:06:38 PM "main", mod_spec)
4/28/2021 10:06:38 PM File "/opt/conda/lib/python3.7/runpy.py", line 85, in _run_code
4/28/2021 10:06:38 PM exec(code, run_globals)
4/28/2021 10:06:38 PM File "/opt/conda/lib/python3.7/site-packages/torch/distributed/launch.py", line 261, in
4/28/2021 10:06:38 PM main()
4/28/2021 10:06:38 PM File "/opt/conda/lib/python3.7/site-packages/torch/distributed/launch.py", line 257, in main
4/28/2021 10:06:38 PM cmd=cmd)
4/28/2021 10:06:38 PM subprocess.CalledProcessError: Command '['/opt/conda/bin/python', '-u', './tools/train.py', '--local_rank=3', 'configs/pointpillars/hv_pointpillars_secfpn_6x8_160e_kitti-3d-car.py', '--launcher', 'pytorch']' returned non-zero exit status 1.
4/28/2021 10:06:38 PM *****************************************
4/28/2021 10:06:38 PM Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.

Could you please give me some suggestion?
Sincerely thanks!

[Wuziyi616]

I am not very familiar with docker. Can you run python mmdet3d/utils/collect_env.py to let us know your environment settings? And can you re-compile mmdet3d (I guess remove this one and re-install another) and show us the output information?

[Dinghow]

I am not very familiar with docker. Can you run python mmdet3d/utils/collect_env.py to let us know your environment settings? And can you re-compile mmdet3d (I guess remove this one and re-install another) and show us the output information?

Sure, the environment in my docker:

sys.platform: linux
4/29/2021 3:23:47 PM Python: 3.7.7 (default, May 7 2020, 21:25:33) [GCC 7.3.0]
4/29/2021 3:23:47 PM CUDA available: True
4/29/2021 3:23:47 PM GPU 0: Tesla V100-SXM2-32GB
4/29/2021 3:23:47 PM CUDA_HOME: /usr/local/cuda
4/29/2021 3:23:47 PM NVCC: Cuda compilation tools, release 10.1, V10.1.243
4/29/2021 3:23:47 PM GCC: gcc (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0
4/29/2021 3:23:47 PM PyTorch: 1.6.0
4/29/2021 3:23:47 PM PyTorch compiling details: PyTorch built with:
4/29/2021 3:23:47 PM - GCC 7.3
4/29/2021 3:23:47 PM - C++ Version: 201402
4/29/2021 3:23:47 PM - Intel(R) Math Kernel Library Version 2020.0.1 Product Build 20200208 for Intel(R) 64 architecture applications
4/29/2021 3:23:47 PM - Intel(R) MKL-DNN v1.5.0 (Git Hash e2ac1fac44c5078ca927cb9b90e1b3066a0b2ed0)
4/29/2021 3:23:47 PM - OpenMP 201511 (a.k.a. OpenMP 4.5)
4/29/2021 3:23:47 PM - NNPACK is enabled
4/29/2021 3:23:47 PM - CPU capability usage: AVX2
4/29/2021 3:23:47 PM - CUDA Runtime 10.1
4/29/2021 3:23:47 PM - NVCC architecture flags: -gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75;-gencode;arch=compute_37,code=compute_37
4/29/2021 3:23:47 PM - CuDNN 7.6.3
4/29/2021 3:23:47 PM - Magma 2.5.2
4/29/2021 3:23:47 PM - Build settings: BLAS=MKL, BUILD_TYPE=Release, CXX_FLAGS= -Wno-deprecated -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -fopenmp -DNDEBUG -DUSE_FBGEMM -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DUSE_XNNPACK -DUSE_VULKAN_WRAPPER -O2 -fPIC -Wno-narrowing -Wall -Wextra -Werror=return-type -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-sign-compare -Wno-unused-parameter -Wno-unused-variable -Wno-unused-function -Wno-unused-result -Wno-unused-local-typedefs -Wno-strict-overflow -Wno-strict-aliasing -Wno-error=deprecated-declarations -Wno-stringop-overflow -Wno-error=pedantic -Wno-error=redundant-decls -Wno-error=old-style-cast -fdiagnostics-color=always -faligned-new -Wno-unused-but-set-variable -Wno-maybe-uninitialized -fno-math-errno -fno-trapping-math -Werror=format -Wno-stringop-overflow, PERF_WITH_AVX=1, PERF_WITH_AVX2=1, PERF_WITH_AVX512=1, USE_CUDA=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=ON, USE_MKLDNN=ON, USE_MPI=OFF, USE_NCCL=ON, USE_NNPACK=ON, USE_OPENMP=ON, USE_STATIC_DISPATCH=OFF,
4/29/2021 3:23:47 PM
4/29/2021 3:23:47 PM TorchVision: 0.7.0
4/29/2021 3:23:47 PM OpenCV: 4.5.1
4/29/2021 3:23:47 PM MMCV: 1.3.1
4/29/2021 3:23:47 PM MMCV Compiler: GCC 7.3
4/29/2021 3:23:47 PM MMCV CUDA Compiler: 10.1
4/29/2021 3:23:47 PM MMDetection: 2.11.0
4/29/2021 3:23:47 PM MMDetection3D: 0.12.0+f7356f4

Thanks for your immediate reply, and I wonder if you update your Dockerfile in time (I notice the last update is 7 month ago), maybe it's out of date? Have you validate the docker environment?

Sincerely thanks!

[Dinghow]

The issue is closed, since we find the official Dockerfile is right, and it happens due to our configuration of our train platform.

Thanks for your help!

[zhyever]

The issue is closed, since we find the official Dockerfile is right, and it happens due to our configuration of our train platform.

Thanks for your help!

hi, I meet the same problem. Can you share the configuration?

[Dinghow]

The issue is closed, since we find the official Dockerfile is right, and it happens due to our configuration of our train platform.
Thanks for your help!

hi, I meet the same problem. Can you share the configuration?

Hello, I met this problem because we deployed a docker environment with mmdet3d offical dockerfile, but we run the code mounted on docker without building, and we solve it by building mounted mmdet3d.