Activity recognition python deepstream implementation

[shrihari]

Hi,

The following is our system configurations,
Deepstream version 6.2
python version 3.8.10
devices : T4(tesla) dGPU
tensorRT version: 8.5.2
pyds version: 1.1.6

We are tyring to implement 3d action recognition model which is available in c++ implementation in Python. We found that some of the APIs are not available for python bindings.

In pgie probe callback function:

GstNvDsPreProcessBatchMeta *preprocess_batchmeta =
          (GstNvDsPreProcessBatchMeta *)(user_meta->user_meta_data);

we couldnt find the python api/method for GstNvDsPreProcessBatchMeta. Because of this we couldn’t proceed further to extract the roi _meta information. Is there any alternate approach for 3d action recognition implementation in deepstream python or are we missing anything?

the c++ code which we are refering to is sources/apps/sample_apps/deepstream-3d-action-recognition

@Fiona.Chen
@yuweiw

Thanks in advance!

[yuweiw]

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

You can try to bind that by yourself. We provide detailed step-by-step instructions. CUSTOMUSERMETAGUIDE.