Spec file related errors for PointPillars

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

Please provide the following information when requesting support.

• Network Type (PointPillars)
• TLT Version (4.0.1)
• Training spec file (pointpillars.yaml from getting_started_v4.0.1)
• How to reproduce the issue?

Generate dataset statistics for data augmentation

!tao pointpillars dataset_convert -e $SPECS_DIR/pointpillars.yaml

the error messages are as below:
2023-09-25 08:43:43,098 [INFO] root: Registry: [‘nvcr.io’]
2023-09-25 08:43:43,167 [INFO] tlt.components.instance_handler.local_instance: Running command in container: nvcr.io/nvidia/tao/tao-toolkit:4.0.0-pyt
2023-09-25 08:43:43,231 [WARNING] tlt.components.docker_handler.docker_handler:
Docker will run the commands as root. If you would like to retain your
local host permissions, please add the “user”:“UID:GID” in the
DockerOptions portion of the “/home/avresearch/.tao_mounts.json” file. You can obtain your
users UID and GID by using the “id -u” and “id -g” commands on the
terminal.
python /opt/conda/lib/python3.8/site-packages/nvidia_tao_pytorch/pointcloud/pointpillars/scripts/dataset_convert.py --cfg_file /home/avresearch/getting_started_v4.0.1/notebooks/tao_launcher_starter_kit/pointpillars/specs/pointpillars.yaml
Traceback (most recent call last):
File “</opt/conda/lib/python3.8/site-packages/nvidia_tao_pytorch/pointcloud/pointpillars/scripts/dataset_convert.py>”, line 3, in **
File “”, line 22, in
AttributeError: ‘EasyDict’ object has no attribute ‘CLASS_NAMES’
2023-09-25 08:43:48,732 [INFO] tlt.components.docker_handler.docker_handler: Stopping container.

It is not clear for me to figure out where to start. Can you provide any solutions for this? I do not get why it says a missing attribute. If you need more information to help me, please let me know. Thank you!

2

Please double check the dataset. Also, suggest you to run with TAO 5.0.
The notebook: https://github.com/NVIDIA/tao_tutorials/blob/main/notebooks/tao_launcher_starter_kit/pointpillars/pointpillars.ipynb
The source code: https://github.com/NVIDIA/tao_pytorch_backend/tree/main/nvidia_tao_pytorch/pointcloud/pointpillars/

3

Thank you for your reply! I’ve used the KITTI dataset following the notebook instructions. Can you please provide more ideas on what to look into within the dataset considering the error messages?

4

Can you share your yaml file?

5

Since you are using TAO 4.0.1, please download the 4.0.1 notebooks.

You can find that below in the example yaml file.

CLASS_NAMES: ['Car', 'Pedestrian', 'Cyclist']
6

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.