enazoe / yolo-tensorrt

TensorRT8.Support Yolov5n,s,m,l,x .darknet -> tensorrt. Yolov4 Yolov3 use raw darknet *.weights and *.cfg fils. If the wrapper is useful to you,please Star it.

MIT License

1.19k stars 316 forks source link

这个calibration_image_list_file_txt是干嘛用的，不设置可以吗？ #127

Closed xinsuinizhuan closed 3 years ago

xinsuinizhuan commented 3 years ago

std::string calibration_image_list_file_txt = "configs/calibration_images.txt";

我看测试例子里面，就Config config_v4没设置，其他都设置了。其他不设置可以吗，是干嘛用的？

enazoe commented 3 years ago

INT8模式较准用的

xinsuinizhuan commented 3 years ago

有校准图片吗？校准是怎么个原理，为啥校准，比较小白，希望指导？

enazoe commented 3 years ago

8

xinsuinizhuan commented 3 years ago

8

感谢。

Lenan22 commented 1 year ago

Please refer to our open source quantization tool ppq, the quantization effect is better than the calibration of tensorrt, if you encounter issues, we can help you solve them. https://github.com/openppl-public/ppq/blob/master/md_doc/deploy_trt_by_OnnxParser.md