ONNXToTensorRT

C++ and Python convert ONNX to TensorRT

prerequisite

Please confirm that you have configured CUDA, CUDNN, TensorRT

quantization

FP32
FP16
INT8

Tested yolov5, yolov6, yolov7, yolov8 conversion successfully

yolov5: https://github.com/ultralytics/yolov5
yolov6: https://github.com/meituan/YOLOv6
yolov7: https://github.com/WongKinYiu/yolov7
yolov8: https://github.com/ultralytics/ultralytics

Note: If you use INT8 quantization, you need an additional file
C++：calibrator.cpp & calibrator.h
Python: calibrator.py

Run

python main.py