How to use pytorch model that generates heatmap as output in deepstream?

[user133662]

• Hardware Platform (Jetson / GPU)
Jetson Orin
• DeepStream Version
6.1.1
• JetPack Version (valid for Jetson only)
5.0.2
• TensorRT Version
8.4.1-1+cuda11.4

I have a pytorch model that count crowds and gives as an output a heatmap which then can be used to count the crowd. I want to be able to run this model in deepstream. I understand that the output is different from models that output bounding boxes.
How can I use this type of model in deepstream?

[Fiona.Chen]

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

@user133662

I understand that the output is different from models that output bounding boxes.
How can I use this type of model in deepstream?

We define the model which output bboxes as the detection model or detector.

We also have samples for the models which are not detectors. You need to customize the output layer parser and output meta data. Please refer to deepstream_tao_apps/apps/tao_others/deepstream-bodypose2d-app at master · NVIDIA-AI-IOT/deepstream_tao_apps (github.com)