I am using FPGA AI Suite to run crossroad_camera_demo in OpenVINO.
The models I use are:
person-vehicle-bike-detection-crossroad-1016
person-attributes-recognition-crossroad-0230
person-reidentification-retail-0288
The device of person-vehicle-bike-detection-crossroad-1016 is HETERO:FPGA,CPU, and the devices of person-attributes-recognition-crossroad-0230 and person-reidentification-retail-0288 are CPU.
If all three models are run on the CPU, the average frame rate is around 45
If the main recognition model 1016 is run through HETERP:FPGA,CPU, and the other two are run through the CPU, the frame rate will not exceed 20
But the person-vehicle-bike-detection-crossroad-1016 model is not divided into many subgraphs. According to the message printed in the terminal: The input graph is split into 2 subgraph(s), CPU 1 FPGA 1.
I want to know where else will affect the frame rate of the model? And how can I improve the frame rate?
Any help would be greatly appreciated, thank you!
