cong

Hi, there, according to the error message _PyCUDA ERROR_, you can use `Context.pop()` to cleanup the context stack.

Please carefully check whether the type and dimension information of the input data is correct.

I have uploaded this 'libmyplugins.so' file, please test again.

> > I have uploaded this 'libmyplugins.so' file, please test again. > > 环境为Ubuntu18.04 with CUDA 10.2 and cuDNN 8.0.0 TensorRT jetpack自带 PyTorch 1.8.0 and TorchVision 0.9.0 OpenCV-Python 4.5 pycuda...

> > > > I have uploaded this 'libmyplugins.so' file, please test again. > > > > > > > > > 环境为Ubuntu18.04 with CUDA 10.2 and cuDNN 8.0.0 TensorRT...

#5 具体可以参考这个issue，就目前来看本项目在Nano上效果不是很好。

已上传

> > 已上传 > > 请问这个文件是怎么生成的 在生成 yolov5.engine 的时候会相应生成这个文件

> 作者您好，我想问一下这个yolov5+deepsort的模型在转换完.engine之后在在板子上跑也是有跟踪效果的吗？ 您好！是的，在板子上运行的效果是一样的，只是板子资源有限且架构不同会有一定的时效性、适配性影响。

> 您好，在[https://github.com/cong/yolov5_deepsort_tensorrt_cpp](https://github.com/cong/yolov5_deepsort_tensorrt_cpp)这个repo中提供了在x86平台下的测试用例，暂时还没有提供在板子上的测试用例。