How to modify deepstream_test3_app.c to recognize our own objects?

Dear all,

How to modify deepstream_test1_app.c to recognize our own objects? Such as: people head, people fist, fish, window, etc?

How to learn this PGIE_CLASS from the start? Or how do we can create object recognition from the scratch?

#define PGIE_CLASS_ID_VEHICLE 0
#define PGIE_CLASS_ID_PERSON 2

Thank you very much in advance.

Warmest Regards,
Suryadi

Hi,

Do you retrain a model for your use case?
If not, it’s recommended to take this tutorial a look first.
https://docs.nvidia.com/deeplearning/digits/index.html

After having your own model, just update the configure file based on the model should be enough.
https://docs.nvidia.com/metropolis/deepstream/4.0/dev-guide/index.html#page/DeepStream_Development_Guide%2Fdeepstream_custom_model.html%23

Thanks.