Where is a place to start learning PGIE thing?

Dear all,

I try to learn sources/apps/sample_apps/deepstream-test2/README
In there, so many jargons that I did not understand, such as:
pgie, sgie1, sgie2, sgie3, PGIE_CLASS_ID_VEHICLE, etc.

What is pgie stands for? primary gstreamer i… e…?

#define PGIE_CLASS_ID_CAR 0
#define PGIE_CLASS_ID_BICYCLE 1
#define PGIE_CLASS_ID_PERSON 2
#define PGIE_CLASS_ID_ROADSIGN 3

How to specify that Car Color[2] is brown? How to create cyan as the 12th color?
How to create tesla as the 21st car make?

dstest2_pgie_config.txt --> Primary_Detector
Car
Bicycle
Person
Roadsign

dstest2_pgie_config.txt --> Secondary_CarColor
black;blue;brown;gold;green;grey;maroon;orange;red;silver;white;yellow

dstest2_sgie1_config.txt --> Secondary_CarMake
acura;audi;bmw;chevrolet;chrysler;dodge;ford;gmc;honda;hyundai;infiniti;jeep;kia;lexus;mazda;mercedes;nissan;subaru;toyota;volkswagen

dstest2_sgie3_config.txt --> Secondary_VehicleTypes
coupe;largevehicle;sedan;suv;truck;van

Thank you very much in advance.

Warmest Regards,
Suryadi

Hi,

You can find some information in the README file.

This sample run one detector and three classifiers.
These configure are for the inference engine separately;

dstest2_pgie_config.txt    <- primary-gie
dstest2_sgie1_config.txt   <- 1st secondary-gie
dstest2_sgie2_config.txt   <- 2nd secondary-gie
dstest2_sgie3_config.txt   <- 3rd secondary-gie

To update the class output, you will need to re-train the model for your own data.
Ex. brown color or car make

Thanks.

Hi,
For more information, please also check https://docs.nvidia.com/metropolis/deepstream/dev-guide/index.html