Part Number: TDA4VM
Now I am trying to implement pipeline on tda4. For example now I have three modules image_preprocess, inference and post_process. image_preprocess is implemented on c66, inference implemented on c71, post_process implemented on a72. The inference use tivxTIDLNode, tivxTIDLNode input parameter and output parameter are set as graph parameters.
I divide the three modules into two parts, preprocess and inference as first part, postprocess as second part, that is the main thread run preprocess and inference, start an new thread run postprocess. As tivxTIDLNode input and output are set graph parameters, I want to use vxGraphParameterEnqueueReadyRef to enqueue input in main thread; the new thread use vxGraphParameterDequeueDoneRef to dequeue output and use output in postprocess. Is it possible?