add Dockerfile

[towardsautonomy]

No description provided.

[YoushaaMurhij]

Hi, thanks for providing this container. How to activate the environment inside the container? no working for me. Can we get rid of conda?

[towardsautonomy]

Hi, thanks for providing this container. How to activate the environment inside the container? no working for me. Can we get rid of conda?

You don't need to use conda inside docker. You can directly run the python scripts, e.g. python tools/create_data.py nuscenes --root-path ./data/nuscenes --out-dir ./data/nuscenes --extra-tag nuscenes --version v1.0 --canbus ./data

[mimichu]

Hey, thanks a lot for sharing your docker file!!!

For some reason, the docker container I built with this (with no errors during the build) is missing pip and torch, which should have been installed by the dockerfile. I would really appreciate it if you have insights into what might went wrong, thank you! So this is what I did:

1. After building the docker file via command:
   docker build --rm -f Dockerfile -t <image_name> .
2. Running the docker via command:
   docker run -it --gpus all <image_name>
3. I tried to train the model using the provided command from BEVFormer instruction:
   ./tools/dist_train.sh ./projects/configs/bevformer/bevformer_base.py 8
   It returns:

   /usr/bin/python3: Error while finding module specification for 'torch.distributed.launch' (ModuleNotFoundError: No module named 'torch')

4. Then I tried to pip install torch, and it returns:

   bash: pip: command not found

[yiakwy-xpu-ml-framework-team]

Could you update gcc, cuda to gcc-9(10, 11 are also welcome), details can be found in TensorRT OSS repository docker file.

gcc-6 is definitely not the right version for Ubuntu-18.04, Ubuntu-20.04 and Ubuntu-22.04 and should be not used for the default version of tool chain.

Building mmdetection3d should use this sentence (otherwise will fail):

python -m pip install --no-cache-dir -e .

[pisquilo]

Thank you so much for this docker file, really helped me.

Due to some changes since the dockerfile was created, there was some packages missing.

Here is what I added to the docker file:

install aditional libraries

RUN pip install fvcore==0.1.5.post20221221
RUN python -m pip install 'git+https://github.com/facebookresearch/detectron2.git'
RUN pip install seaborn==0.12.0

[yiakwy-xpu-ml-framework-team]

您好，邮箱主人会认真阅读！谢谢关注/