Tao toolkit on aws cloud

Hi Team,

we followed following instruction on installation of tao toolkit on AWS instance. however after installation when we run tao info… we are not getting mask_rcnn option withing the option.

https://docs.nvidia.com/tao/tao-toolkit/text/running_in_cloud/running_tao_toolkit_on_aws.html

Configuration of the TAO Toolkit Instance

task_group: [‘model’, ‘dataset’, ‘deploy’]

format_version: 3.0

toolkit_version: 5.0.0

published_date: 07/14/2023

how do we configure tao toolkit to get this option?

On this page Running TAO Toolkit on an AWS VM - NVIDIA Docs ,

Below is an old version. Sorry for not updating it.

wget --content-disposition https://api.ngc.nvidia.com/v2/resources/nvidia/tao/cv_samples/versions/v1.2.0/zip -O cv_samples_v1.2.0.zip
unzip -u cv_samples_v1.2.0.zip -d ./cv_samples_v1.2.0 && cd ./cv_samples_v1.2.0

Please refer to https://docs.nvidia.com/tao/tao-toolkit/text/tao_toolkit_quick_start_guide.html???
Below will download latest 5.0 notebook.

wget --content-disposition https://api.ngc.nvidia.com/v2/resources/nvidia/tao/tao-getting-started/versions/5.0.0/zip -O getting_started_v5.0.0.zip
unzip -u getting_started_v5.0.0.zip -d ./getting_started_v5.0.0 && rm -rf getting_started_v5.0.0.zip && cd ./getting_started_v5.0.0

1 Like

Thank you for the above update. it’s working now.

we are trying to run an Action recognition using TAO ActionRecognitionNet … first testing with a sample dataset to make sure everything works… however, stuck on the following step :

any help why its giving this error?

Did you follow the default notebook? For this error, please double check the dataset. Please try to run the default notebook and default dataset.

Yes, there is no modification done we just followed the default notebook and it downloaded the default dataset. however, it can not convert video to a png file it gives the above error.

Please try to check why the shell script failed to run well against the dataset.
The source code is in

I recall that other user did not meet this kind of issue.

I think “preprocess README” is not expected. Maybe it is related to the path.

we modify the path and then it started working…

import os
os.chdir(‘tao_toolkit_recipes/tao_action_recognition/data_generation’)

import os
current_working_directory = os.getcwd()
print(current_working_directory)

so what could be the error in the default notebook.?

Could you save the notebook as .html file and share with us?
Thanks.

After modification we still stuck on training error.

actionrecognitionnet (1).html (832 KB)

Please delete https://github.com/NVIDIA/tao_tutorials/blob/main/notebooks/tao_launcher_starter_kit/action_recognition_net/specs/train_rgb_3d_finetune.yaml#L1 and retry.

Thank you … however we are using this file to train model… following tutorial from

https://www.youtube.com/watch?v=9cSyNtpIqLc

print(“Train RGB only model with PTM”)

!tao model action_recognition train \

-e $SPECS_DIR/train_rgb_3d_finetune.yaml \

#-r $RESULTS_DIR/rgb_3d_ptm \

results_dir=$RESULTS_DIR/rgb_3d_ptm \

-k $KEY \

model.rgb_pretrained_model_path=$RESULTS_DIR/pretrained/actionrecognitionnet_vtrainable_v1.0/resnet18_3d_rgb_hmdb5_32.tlt \

model.rgb_pretrained_num_classes=5

should we be using different file ?

Can you share the yaml file?
This file is actually output_dir: /results/rgb_3d_ptm.

For workaround, please change
output_dir: /results/rgb_3d_ptm

to
results_dir: /results/rgb_3d_ptm

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