tothedistance

@NVukobrat thank you very much

create_train_data.py show how the plate and label are generated. 11,0 are indexes of array consists of chinese province chars and alphanumeric chars

换背景文件夹里的背景图片。

> > > > 先使用这个[https://github.com/daquexian/onnx-simplifier，可以去掉这些层](https://github.com/daquexian/onnx-simplifier%EF%BC%8C%E5%8F%AF%E4%BB%A5%E5%8E%BB%E6%8E%89%E8%BF%99%E4%BA%9B%E5%B1%82) > > 您好，我该怎么用onnx-simplifier去掉reshape节点呢 python -m onnxsim onnx模型名称 yolov5s-simple.onnx 得到最终简化后的onnx模型 https://github.com/Wulingtian/yolov5_onnx2caffe

detect层里算坐标里面有shape。但一般移植都是cpu算这部分，simplifier刚好去掉了。估计原理是不支持的就去掉，但感觉不稳定。

i guess we have to put all header under /usr/include. for my company's machine, all header are under /usr/include/rockchip. and it seems the gcc will not go down recursively.

for those who install boost using vcpkg, dont forget to add the toolchain. while the error seems to find the boost, it's file structure is totally different, thus producing confusing...

@YXL76 set buildDirectory to "${sourceDirectory}/build" and delete all existing build leftover and run command "reset cmake tools" might help.