Try to run ssd_mobilenet_v1_trt model but crashed

Robotics & Edge Computing Jetson & Embedded Systems Jetson Nano
1

I optimized the ssd_mobilenet_v1 model to ssd_mobilenet_v1_trt model by follow this link GitHub - NVIDIA-AI-IOT/tf_trt_models: TensorFlow models accelerated with NVIDIA TensorRT.

But when I run the sample code in Juston Nano’s jupyter notebook, it told me that kernel has been crashed. Why?

2

Hi,

You can check this topic for model ssd_mobilenet_v1 to TensorRT sample:
[url]https://github.com/AastaNV/TRT_object_detection[/url]

Thanks.

3

Thank you.

I’ll try this repo.

And there are the update of my issue. I try to run the origin code in Terminal, there are the error message.

Default Logger The engine plan file is incompatible with this version of TensorRT, expecting 5.0.6.3 got 5.0.2.6, please rebuild

It told me that the version of TensorRT was incompatible, acturally I optimized the model in PC which has GTX 1066 TensorRT 5.0.2.6, and the Jeston Nano’s TensorRT version is 5.0.6.3.

How can I install the TensorRT 5.0.6.3 in my PC?

4

Hi,

I try to follow the repo but stopped in the environment, this repo use pycuda, but when I try to install the pycuda by pip3, the wheel told me that cuda.h: No such file or directory.

In file included from src/cpp/cuda.cpp:1:0:
      src/cpp/cuda.hpp:14:10: fatal error: cuda.h: No such file or directory
      #include <cuda.h>
      compilation terminated.
      error: command 'aarch64-linux-gnu-gcc' failed with exit status 1
5

I found the solution in here Installing PyCUDA on Ubuntu Linux - Andreas Klöckner's Former Wiki

export CUDA_ROOT=/usr/local/cuda
export PATH=$PATH:$CUDA_ROOT/bin
6

Thank you! I succeeded！

7

Good to hear this : )