Deployment and accuracy assessment of classification models on Jetson Orin NANO with TensorRT
- 编程语言 Programming Language:C++
- 额外库 Extra Libraries:
- TensorRT (C++)
- nlohmann's json
本代码需要读取TensorRT engine文件(.trt),代码本身不包含生成engine的部分,请使用TensorRT的trtexec
工具进行快速转换,或自行搜索其他使用代码生成engine的脚本。详情请参考TensorRT的官方文档(太长了我也没全看)。
This code needs to read a TensorRT engine file, whose extension is usually ".trt", and the code doesn't generates one, so please use trtexec
from TensorRT for simple engine generation, or search for some script that generates engine files on your own. For details, see official documents from TensorRT. (To your amusement I didn't read it to the end cuz it's too long ... orz)
代码中包含了必要的英文注释,如有疑问欢迎交流。
The code contains necessary comments in it, for which if you have any question, discussions are welcome.
本代码并未使用对象式的编程,这是不好的。但是目前还有其他工作要做,所以就先不改了=。= 大家可以自行参考或修改。但请注意遵守代码许可。
The code was not written in OOP style, which is not very elegant. However the styling thing is not very urgent for me so... ¯\_(ツ)_/¯ You can copy or modify the code by yourself as you like it. But please pay attention to the licenses.
本代码参考了下面几篇博客,非常感谢这些乐于分享的前辈们:
References are listed below. Sincere gratitude to these helping seniors: