I need a small clarification. My product application is like reading ip camera streaming and applying inferencing model and based on the prediction i need to send custom meta data to broker. (Eg: number of person detected, time, customer details,camera ID,only when person detected in specific time etc). I have gone through graph composer. It is fine for me to proceed with. but i dont know whether we can customize the meta data which is coming out from inference model or we cannot customize and add own logics to output of inference prediction?
if it is not feasible then i have to move with tensorrt and opencv based python implementation for my purpose.
DeepStream is just a SDK. There is no limitation of you to use other APIs. You can get meta data with DeepStream API and send the data to othere places with other API
Isn’t Graph Composer just at the very beginning? Maybe there will be an option added in later versions? I really like the idea of this GUI solution for building AIs