LibTorch-YoLoV5-train-detection
use c++ train yolov5
Just for learning!!!
Just for learning!!!
Just for learning!!!
Dependencies
Windows10
Qt5.13.2(kit MSVC2017 64bit)
CUDA 11.1
OpenCV 4.5.2
LibTorch 1.8.2
SetUp
1. take your libtorch/opencv *.dll files to "./bin/realease" or "./bin/debug"
2. uncheck "Shadow build" in QtCreator
3. get TorchScript pt file from https://github.com/ultralytics/yolov5/export.py
or my release version: https://github.com/jedi007/LibTorch-YoLoV5-train-detection/releases/download/yolov5s.torchscript.cuda.unfuse.pt/yolov5s.torchscript.cuda.unfuse.zip
Be careful the TorchScript pt file must export as "fuse=False"
References
https://github.com/ultralytics/yolov5
https://github.com/yasenh/libtorch-yolov5