I’m trained a yolov3 model with my datasets, I want used deepstream test this model of tensorrt int8, but the result is too bad,so I don’t know the result cause by tensorrt calibration, so I want test again with a custom yolov3 tensorrt calibration, help me how to create it, thanks very much.
Hi,
Please check following document for the INT8 calibration:
C++:
https://github.com/NVIDIA/TensorRT/tree/release/6.0/samples/opensource/sampleINT8API
Python example:
Thanks.