No action recogniton net model classes identified

[laura.perez]

I have installed Deepstream 6.2 on a Jetson with Ubuntu 20.04.
I want to run the action recognition model. When I run the command:

deepstream-3d-action-recognition -c deepstream_action_recognition_config.txt

I get different warnings, one of them, that I don’t know how to solve is the following:
WARNING from element primary-nvinference-engine: nvinfer could not find input layer with name = input_1

Warning: nvinfer could not find input layer with name = input_1

How can I find out the name of the input layer?

On the other hand, in the inference file I don’t know which files should be put in the following variables:
tlt-encoded-model=./resnet18_3d_of_hmdb5_32_a100.etlt
tlt-model-key=nvidia_tao
model-engine-file=/triton_model_repo/resnet18_3d_rgb/resnet18_3d_rgb.engine

As far as I have understood in encoded model you have to put one of the files that you download in the “Action Recognition Net” page, but what to put in “model-engine-file”?

Thanks!

[fanzh]

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks

@laura.perez

How can I find out the name of the input layer?

I can’t reproduce this issue, please share your configuration and logs.
log.txt (7.6 KB)

@laura.perez

As far as I have understood in encoded model you have to put one of the files that you download in the “Action Recognition Net” page, but what to put in “model-engine-file”?

if setting model-engine-file, the application will load that engine directly instead of generating a new one, please ignore it at the first run.